blender/extern/libmv/libmv-capi.cpp

978 lines
31 KiB
C++
Raw Normal View History

Camera tracking integration =========================== Commiting camera tracking integration gsoc project into trunk. This commit includes: - Bundled version of libmv library (with some changes against official repo, re-sync with libmv repo a bit later) - New datatype ID called MovieClip which is optimized to work with movie clips (both of movie files and image sequences) and doing camera/motion tracking operations. - New editor called Clip Editor which is currently used for motion/tracking stuff only, but which can be easily extended to work with masks too. This editor supports: * Loading movie files/image sequences * Build proxies with different size for loaded movie clip, also supports building undistorted proxies to increase speed of playback in undistorted mode. * Manual lens distortion mode calibration using grid and grease pencil * Supervised 2D tracking using two different algorithms KLT and SAD. * Basic algorithm for feature detection * Camera motion solving. scene orientation - New constraints to "link" scene objects with solved motions from clip: * Follow Track (make object follow 2D motion of track with given name or parent object to reconstructed 3D position of track) * Camera Solver to make camera moving in the same way as reconstructed camera This commit NOT includes changes from tomato branch: - New nodes (they'll be commited as separated patch) - Automatic image offset guessing for image input node and image editor (need to do more tests and gather more feedback) - Code cleanup in libmv-capi. It's not so critical cleanup, just increasing readability and understanadability of code. Better to make this chaneg when Keir will finish his current patch. More details about this project can be found on this page: http://wiki.blender.org/index.php/User:Nazg-gul/GSoC-2011 Further development of small features would be done in trunk, bigger/experimental features would first be implemented in tomato branch.
2011-11-07 12:55:18 +00:00
/*
* ***** BEGIN GPL LICENSE BLOCK *****
*
* This program is free software; you can redistribute it and/or
* modify it under the terms of the GNU General Public License
* as published by the Free Software Foundation; either version 2
* of the License, or (at your option) any later version.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public License
* along with this program; if not, write to the Free Software Foundation,
* Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
*
* The Original Code is Copyright (C) 2011 Blender Foundation.
* All rights reserved.
*
* Contributor(s): Blender Foundation,
* Sergey Sharybin
*
* ***** END GPL LICENSE BLOCK *****
*/
/* define this to generate PNG images with content of search areas
tracking between which failed */
#undef DUMP_FAILURE
Planar tracking support for motion tracking =========================================== Major list of changes done in tomato branch: - Add a planar tracking implementation to libmv This adds a new planar tracking implementation to libmv. The tracker is based on Ceres[1], the new nonlinear minimizer that myself and Sameer released from Google as open source. Since the motion model is more involved, the interface is different than the RegionTracker interface used previously in Blender. The start of a C API in libmv-capi.{cpp,h} is also included. - Migrate from pat_{min,max} for markers to 4 corners representation Convert markers in the movie clip editor / 2D tracker from using pat_min and pat_max notation to using the a more general, 4-corner representation. There is still considerable porting work to do; in particular sliding from preview widget does not work correct for rotated markers. All other areas should be ported to new representation: * Added support of sliding individual corners. LMB slide + Ctrl would scale the whole pattern * S would scale the whole marker, S-S would scale pattern only * Added support of marker's rotation which is currently rotates only patterns around their centers or all markers around median, Rotation or other non-translation/scaling transformation of search area doesn't make sense. * Track Preview widget would display transformed pattern which libmv actually operates with. - "Efficient Second-order Minimization" for the planar tracker This implements the "Efficient Second-order Minimization" scheme, as supported by the existing translation tracker. This increases the amount of per-iteration work, but decreases the number of iterations required to converge and also increases the size of the basin of attraction for the optimization. - Remove the use of the legacy RegionTracker API from Blender, and replaces it with the new TrackRegion API. This also adds several features to the planar tracker in libmv: * Do a brute-force initialization of tracking similar to "Hybrid" mode in the stable release, but using all floats. This is slower but more accurate. It is still necessary to evaluate if the performance loss is worth it. In particular, this change is necessary to support high bit depth imagery. * Add support for masks over the search window. This is a step towards supporting user-defined tracker masks. The tracker masks will make it easy for users to make a mask for e.g. a ball. Not exposed into interface yet/ * Add Pearson product moment correlation coefficient checking (aka "Correlation" in the UI. This causes tracking failure if the tracked patch is not linearly related to the template. * Add support for warping a few points in addition to the supplied points. This is useful because the tracking code deliberately does not expose the underlying warp representation. Instead, warps are specified in an aparametric way via the correspondences. - Replace the old style tracker configuration panel with the new planar tracking panel. From a users perspective, this means: * The old "tracking algorithm" picker is gone. There is only 1 algorithm now. We may revisit this later, but I would much prefer to have only 1 algorithm. So far no optimization work has been done so the speed is not there yet. * There is now a dropdown to select the motion model. Choices: * Translation * Translation, rotation * Translation, scale * Translation, rotation, scale * Affine * Perspective * The old "Hybrid" mode is gone; instead there is a toggle to enable or disable translation-only tracker initialization. This is the equivalent of the hyrbid mode before, but rewritten to work with the new planar tracking modes. * The pyramid levels setting is gone. At a future date, the planar tracker will decide to use pyramids or not automatically. The pyramid setting was ultimately a mistake; with the brute force initialization it is unnecessary. - Add light-normalized tracking Added the ability to normalize patterns by their average value while tracking, to make them invariant to global illumination changes. Additional details could be found at wiki page [2] [1] http://code.google.com/p/ceres-solver [2] http://wiki.blender.org/index.php/Dev:Ref/Release_Notes/2.64/Motion_Tracker
2012-06-10 15:28:19 +00:00
/* define this to generate PNG images with content of search areas
on every itteration of tracking */
#undef DUMP_ALWAYS
Camera tracking integration =========================== Commiting camera tracking integration gsoc project into trunk. This commit includes: - Bundled version of libmv library (with some changes against official repo, re-sync with libmv repo a bit later) - New datatype ID called MovieClip which is optimized to work with movie clips (both of movie files and image sequences) and doing camera/motion tracking operations. - New editor called Clip Editor which is currently used for motion/tracking stuff only, but which can be easily extended to work with masks too. This editor supports: * Loading movie files/image sequences * Build proxies with different size for loaded movie clip, also supports building undistorted proxies to increase speed of playback in undistorted mode. * Manual lens distortion mode calibration using grid and grease pencil * Supervised 2D tracking using two different algorithms KLT and SAD. * Basic algorithm for feature detection * Camera motion solving. scene orientation - New constraints to "link" scene objects with solved motions from clip: * Follow Track (make object follow 2D motion of track with given name or parent object to reconstructed 3D position of track) * Camera Solver to make camera moving in the same way as reconstructed camera This commit NOT includes changes from tomato branch: - New nodes (they'll be commited as separated patch) - Automatic image offset guessing for image input node and image editor (need to do more tests and gather more feedback) - Code cleanup in libmv-capi. It's not so critical cleanup, just increasing readability and understanadability of code. Better to make this chaneg when Keir will finish his current patch. More details about this project can be found on this page: http://wiki.blender.org/index.php/User:Nazg-gul/GSoC-2011 Further development of small features would be done in trunk, bigger/experimental features would first be implemented in tomato branch.
2011-11-07 12:55:18 +00:00
#include "libmv-capi.h"
#include "libmv/logging/logging.h"
Planar tracking support for motion tracking =========================================== Major list of changes done in tomato branch: - Add a planar tracking implementation to libmv This adds a new planar tracking implementation to libmv. The tracker is based on Ceres[1], the new nonlinear minimizer that myself and Sameer released from Google as open source. Since the motion model is more involved, the interface is different than the RegionTracker interface used previously in Blender. The start of a C API in libmv-capi.{cpp,h} is also included. - Migrate from pat_{min,max} for markers to 4 corners representation Convert markers in the movie clip editor / 2D tracker from using pat_min and pat_max notation to using the a more general, 4-corner representation. There is still considerable porting work to do; in particular sliding from preview widget does not work correct for rotated markers. All other areas should be ported to new representation: * Added support of sliding individual corners. LMB slide + Ctrl would scale the whole pattern * S would scale the whole marker, S-S would scale pattern only * Added support of marker's rotation which is currently rotates only patterns around their centers or all markers around median, Rotation or other non-translation/scaling transformation of search area doesn't make sense. * Track Preview widget would display transformed pattern which libmv actually operates with. - "Efficient Second-order Minimization" for the planar tracker This implements the "Efficient Second-order Minimization" scheme, as supported by the existing translation tracker. This increases the amount of per-iteration work, but decreases the number of iterations required to converge and also increases the size of the basin of attraction for the optimization. - Remove the use of the legacy RegionTracker API from Blender, and replaces it with the new TrackRegion API. This also adds several features to the planar tracker in libmv: * Do a brute-force initialization of tracking similar to "Hybrid" mode in the stable release, but using all floats. This is slower but more accurate. It is still necessary to evaluate if the performance loss is worth it. In particular, this change is necessary to support high bit depth imagery. * Add support for masks over the search window. This is a step towards supporting user-defined tracker masks. The tracker masks will make it easy for users to make a mask for e.g. a ball. Not exposed into interface yet/ * Add Pearson product moment correlation coefficient checking (aka "Correlation" in the UI. This causes tracking failure if the tracked patch is not linearly related to the template. * Add support for warping a few points in addition to the supplied points. This is useful because the tracking code deliberately does not expose the underlying warp representation. Instead, warps are specified in an aparametric way via the correspondences. - Replace the old style tracker configuration panel with the new planar tracking panel. From a users perspective, this means: * The old "tracking algorithm" picker is gone. There is only 1 algorithm now. We may revisit this later, but I would much prefer to have only 1 algorithm. So far no optimization work has been done so the speed is not there yet. * There is now a dropdown to select the motion model. Choices: * Translation * Translation, rotation * Translation, scale * Translation, rotation, scale * Affine * Perspective * The old "Hybrid" mode is gone; instead there is a toggle to enable or disable translation-only tracker initialization. This is the equivalent of the hyrbid mode before, but rewritten to work with the new planar tracking modes. * The pyramid levels setting is gone. At a future date, the planar tracker will decide to use pyramids or not automatically. The pyramid setting was ultimately a mistake; with the brute force initialization it is unnecessary. - Add light-normalized tracking Added the ability to normalize patterns by their average value while tracking, to make them invariant to global illumination changes. Additional details could be found at wiki page [2] [1] http://code.google.com/p/ceres-solver [2] http://wiki.blender.org/index.php/Dev:Ref/Release_Notes/2.64/Motion_Tracker
2012-06-10 15:28:19 +00:00
#include "libmv/tracking/track_region.h"
Camera tracking integration =========================== Commiting camera tracking integration gsoc project into trunk. This commit includes: - Bundled version of libmv library (with some changes against official repo, re-sync with libmv repo a bit later) - New datatype ID called MovieClip which is optimized to work with movie clips (both of movie files and image sequences) and doing camera/motion tracking operations. - New editor called Clip Editor which is currently used for motion/tracking stuff only, but which can be easily extended to work with masks too. This editor supports: * Loading movie files/image sequences * Build proxies with different size for loaded movie clip, also supports building undistorted proxies to increase speed of playback in undistorted mode. * Manual lens distortion mode calibration using grid and grease pencil * Supervised 2D tracking using two different algorithms KLT and SAD. * Basic algorithm for feature detection * Camera motion solving. scene orientation - New constraints to "link" scene objects with solved motions from clip: * Follow Track (make object follow 2D motion of track with given name or parent object to reconstructed 3D position of track) * Camera Solver to make camera moving in the same way as reconstructed camera This commit NOT includes changes from tomato branch: - New nodes (they'll be commited as separated patch) - Automatic image offset guessing for image input node and image editor (need to do more tests and gather more feedback) - Code cleanup in libmv-capi. It's not so critical cleanup, just increasing readability and understanadability of code. Better to make this chaneg when Keir will finish his current patch. More details about this project can be found on this page: http://wiki.blender.org/index.php/User:Nazg-gul/GSoC-2011 Further development of small features would be done in trunk, bigger/experimental features would first be implemented in tomato branch.
2011-11-07 12:55:18 +00:00
#include "libmv/simple_pipeline/callbacks.h"
Camera tracking integration =========================== Commiting camera tracking integration gsoc project into trunk. This commit includes: - Bundled version of libmv library (with some changes against official repo, re-sync with libmv repo a bit later) - New datatype ID called MovieClip which is optimized to work with movie clips (both of movie files and image sequences) and doing camera/motion tracking operations. - New editor called Clip Editor which is currently used for motion/tracking stuff only, but which can be easily extended to work with masks too. This editor supports: * Loading movie files/image sequences * Build proxies with different size for loaded movie clip, also supports building undistorted proxies to increase speed of playback in undistorted mode. * Manual lens distortion mode calibration using grid and grease pencil * Supervised 2D tracking using two different algorithms KLT and SAD. * Basic algorithm for feature detection * Camera motion solving. scene orientation - New constraints to "link" scene objects with solved motions from clip: * Follow Track (make object follow 2D motion of track with given name or parent object to reconstructed 3D position of track) * Camera Solver to make camera moving in the same way as reconstructed camera This commit NOT includes changes from tomato branch: - New nodes (they'll be commited as separated patch) - Automatic image offset guessing for image input node and image editor (need to do more tests and gather more feedback) - Code cleanup in libmv-capi. It's not so critical cleanup, just increasing readability and understanadability of code. Better to make this chaneg when Keir will finish his current patch. More details about this project can be found on this page: http://wiki.blender.org/index.php/User:Nazg-gul/GSoC-2011 Further development of small features would be done in trunk, bigger/experimental features would first be implemented in tomato branch.
2011-11-07 12:55:18 +00:00
#include "libmv/simple_pipeline/tracks.h"
#include "libmv/simple_pipeline/initialize_reconstruction.h"
#include "libmv/simple_pipeline/bundle.h"
#include "libmv/simple_pipeline/detect.h"
#include "libmv/simple_pipeline/pipeline.h"
#include "libmv/simple_pipeline/camera_intrinsics.h"
#include "libmv/simple_pipeline/modal_solver.h"
Camera tracking integration =========================== Commiting camera tracking integration gsoc project into trunk. This commit includes: - Bundled version of libmv library (with some changes against official repo, re-sync with libmv repo a bit later) - New datatype ID called MovieClip which is optimized to work with movie clips (both of movie files and image sequences) and doing camera/motion tracking operations. - New editor called Clip Editor which is currently used for motion/tracking stuff only, but which can be easily extended to work with masks too. This editor supports: * Loading movie files/image sequences * Build proxies with different size for loaded movie clip, also supports building undistorted proxies to increase speed of playback in undistorted mode. * Manual lens distortion mode calibration using grid and grease pencil * Supervised 2D tracking using two different algorithms KLT and SAD. * Basic algorithm for feature detection * Camera motion solving. scene orientation - New constraints to "link" scene objects with solved motions from clip: * Follow Track (make object follow 2D motion of track with given name or parent object to reconstructed 3D position of track) * Camera Solver to make camera moving in the same way as reconstructed camera This commit NOT includes changes from tomato branch: - New nodes (they'll be commited as separated patch) - Automatic image offset guessing for image input node and image editor (need to do more tests and gather more feedback) - Code cleanup in libmv-capi. It's not so critical cleanup, just increasing readability and understanadability of code. Better to make this chaneg when Keir will finish his current patch. More details about this project can be found on this page: http://wiki.blender.org/index.php/User:Nazg-gul/GSoC-2011 Further development of small features would be done in trunk, bigger/experimental features would first be implemented in tomato branch.
2011-11-07 12:55:18 +00:00
#include <stdlib.h>
Assorted camera tracker improvements - Add support for refining the camera's intrinsic parameters during a solve. Currently, refining supports only the following combinations of intrinsic parameters: f f, cx, cy f, cx, cy, k1, k2 f, k1 f, k1, k2 This is not the same as autocalibration, since the user must still make a reasonable initial guess about the focal length and other parameters, whereas true autocalibration would eliminate the need for the user specify intrinsic parameters at all. However, the solver works well with only rough guesses for the focal length, so perhaps full autocalibation is not that important. Adding support for the last two combinations, (f, k1) and (f, k1, k2) required changes to the library libmv depends on for bundle adjustment, SSBA. These changes should get ported upstream not just to libmv but to SSBA as well. - Improved the region of convergence for bundle adjustment by increasing the number of Levenberg-Marquardt iterations from 50 to 500. This way, the solver is able to crawl out of the bad local minima it gets stuck in when changing from, for example, bundling k1 and k2 to just k1 and resetting k2 to 0. - Add several new region tracker implementations. A region tracker is a libmv concept, which refers to tracking a template image pattern through frames. The impact to end users is that tracking should "just work better". I am reserving a more detailed writeup, and maybe a paper, for later. - Other libmv tweaks, such as detecting that a tracker is headed outside of the image bounds. This includes several changes made directly to the libmv extern code rather expecting to get those changes through normal libmv channels, because I, the libmv BDFL, decided it was faster to work on libmv directly in Blender, then later reverse-port the libmv changes from Blender back into libmv trunk. The interesting part is that I added a full Levenberg-Marquardt loop to the region tracking code, which should lead to a more stable solutions. I also added a hacky implementation of "Efficient Second-Order Minimization" for tracking, which works nicely. A more detailed quantitative evaluation will follow. Original patch by Keir, cleaned a bit by myself.
2011-11-14 06:41:23 +00:00
#include <assert.h>
Camera tracking integration =========================== Commiting camera tracking integration gsoc project into trunk. This commit includes: - Bundled version of libmv library (with some changes against official repo, re-sync with libmv repo a bit later) - New datatype ID called MovieClip which is optimized to work with movie clips (both of movie files and image sequences) and doing camera/motion tracking operations. - New editor called Clip Editor which is currently used for motion/tracking stuff only, but which can be easily extended to work with masks too. This editor supports: * Loading movie files/image sequences * Build proxies with different size for loaded movie clip, also supports building undistorted proxies to increase speed of playback in undistorted mode. * Manual lens distortion mode calibration using grid and grease pencil * Supervised 2D tracking using two different algorithms KLT and SAD. * Basic algorithm for feature detection * Camera motion solving. scene orientation - New constraints to "link" scene objects with solved motions from clip: * Follow Track (make object follow 2D motion of track with given name or parent object to reconstructed 3D position of track) * Camera Solver to make camera moving in the same way as reconstructed camera This commit NOT includes changes from tomato branch: - New nodes (they'll be commited as separated patch) - Automatic image offset guessing for image input node and image editor (need to do more tests and gather more feedback) - Code cleanup in libmv-capi. It's not so critical cleanup, just increasing readability and understanadability of code. Better to make this chaneg when Keir will finish his current patch. More details about this project can be found on this page: http://wiki.blender.org/index.php/User:Nazg-gul/GSoC-2011 Further development of small features would be done in trunk, bigger/experimental features would first be implemented in tomato branch.
2011-11-07 12:55:18 +00:00
Planar tracking support for motion tracking =========================================== Major list of changes done in tomato branch: - Add a planar tracking implementation to libmv This adds a new planar tracking implementation to libmv. The tracker is based on Ceres[1], the new nonlinear minimizer that myself and Sameer released from Google as open source. Since the motion model is more involved, the interface is different than the RegionTracker interface used previously in Blender. The start of a C API in libmv-capi.{cpp,h} is also included. - Migrate from pat_{min,max} for markers to 4 corners representation Convert markers in the movie clip editor / 2D tracker from using pat_min and pat_max notation to using the a more general, 4-corner representation. There is still considerable porting work to do; in particular sliding from preview widget does not work correct for rotated markers. All other areas should be ported to new representation: * Added support of sliding individual corners. LMB slide + Ctrl would scale the whole pattern * S would scale the whole marker, S-S would scale pattern only * Added support of marker's rotation which is currently rotates only patterns around their centers or all markers around median, Rotation or other non-translation/scaling transformation of search area doesn't make sense. * Track Preview widget would display transformed pattern which libmv actually operates with. - "Efficient Second-order Minimization" for the planar tracker This implements the "Efficient Second-order Minimization" scheme, as supported by the existing translation tracker. This increases the amount of per-iteration work, but decreases the number of iterations required to converge and also increases the size of the basin of attraction for the optimization. - Remove the use of the legacy RegionTracker API from Blender, and replaces it with the new TrackRegion API. This also adds several features to the planar tracker in libmv: * Do a brute-force initialization of tracking similar to "Hybrid" mode in the stable release, but using all floats. This is slower but more accurate. It is still necessary to evaluate if the performance loss is worth it. In particular, this change is necessary to support high bit depth imagery. * Add support for masks over the search window. This is a step towards supporting user-defined tracker masks. The tracker masks will make it easy for users to make a mask for e.g. a ball. Not exposed into interface yet/ * Add Pearson product moment correlation coefficient checking (aka "Correlation" in the UI. This causes tracking failure if the tracked patch is not linearly related to the template. * Add support for warping a few points in addition to the supplied points. This is useful because the tracking code deliberately does not expose the underlying warp representation. Instead, warps are specified in an aparametric way via the correspondences. - Replace the old style tracker configuration panel with the new planar tracking panel. From a users perspective, this means: * The old "tracking algorithm" picker is gone. There is only 1 algorithm now. We may revisit this later, but I would much prefer to have only 1 algorithm. So far no optimization work has been done so the speed is not there yet. * There is now a dropdown to select the motion model. Choices: * Translation * Translation, rotation * Translation, scale * Translation, rotation, scale * Affine * Perspective * The old "Hybrid" mode is gone; instead there is a toggle to enable or disable translation-only tracker initialization. This is the equivalent of the hyrbid mode before, but rewritten to work with the new planar tracking modes. * The pyramid levels setting is gone. At a future date, the planar tracker will decide to use pyramids or not automatically. The pyramid setting was ultimately a mistake; with the brute force initialization it is unnecessary. - Add light-normalized tracking Added the ability to normalize patterns by their average value while tracking, to make them invariant to global illumination changes. Additional details could be found at wiki page [2] [1] http://code.google.com/p/ceres-solver [2] http://wiki.blender.org/index.php/Dev:Ref/Release_Notes/2.64/Motion_Tracker
2012-06-10 15:28:19 +00:00
#if defined(DUMP_FAILURE) || defined (DUMP_ALWAYS)
Camera tracking integration =========================== Commiting camera tracking integration gsoc project into trunk. This commit includes: - Bundled version of libmv library (with some changes against official repo, re-sync with libmv repo a bit later) - New datatype ID called MovieClip which is optimized to work with movie clips (both of movie files and image sequences) and doing camera/motion tracking operations. - New editor called Clip Editor which is currently used for motion/tracking stuff only, but which can be easily extended to work with masks too. This editor supports: * Loading movie files/image sequences * Build proxies with different size for loaded movie clip, also supports building undistorted proxies to increase speed of playback in undistorted mode. * Manual lens distortion mode calibration using grid and grease pencil * Supervised 2D tracking using two different algorithms KLT and SAD. * Basic algorithm for feature detection * Camera motion solving. scene orientation - New constraints to "link" scene objects with solved motions from clip: * Follow Track (make object follow 2D motion of track with given name or parent object to reconstructed 3D position of track) * Camera Solver to make camera moving in the same way as reconstructed camera This commit NOT includes changes from tomato branch: - New nodes (they'll be commited as separated patch) - Automatic image offset guessing for image input node and image editor (need to do more tests and gather more feedback) - Code cleanup in libmv-capi. It's not so critical cleanup, just increasing readability and understanadability of code. Better to make this chaneg when Keir will finish his current patch. More details about this project can be found on this page: http://wiki.blender.org/index.php/User:Nazg-gul/GSoC-2011 Further development of small features would be done in trunk, bigger/experimental features would first be implemented in tomato branch.
2011-11-07 12:55:18 +00:00
# include <png.h>
#endif
#ifdef _MSC_VER
# define snprintf _snprintf
#endif
typedef struct libmv_Reconstruction {
libmv::EuclideanReconstruction reconstruction;
/* used for per-track average error calculation after reconstruction */
libmv::Tracks tracks;
libmv::CameraIntrinsics intrinsics;
double error;
} libmv_Reconstruction;
typedef struct libmv_Features {
int count, margin;
libmv::Feature *features;
} libmv_Features;
/* ************ Logging ************ */
void libmv_initLogging(const char *argv0)
{
/* Make it so FATAL messages are always print into console */
char severity_fatal[32];
snprintf(severity_fatal, sizeof(severity_fatal), "%d",
google::GLOG_FATAL);
Camera tracking integration =========================== Commiting camera tracking integration gsoc project into trunk. This commit includes: - Bundled version of libmv library (with some changes against official repo, re-sync with libmv repo a bit later) - New datatype ID called MovieClip which is optimized to work with movie clips (both of movie files and image sequences) and doing camera/motion tracking operations. - New editor called Clip Editor which is currently used for motion/tracking stuff only, but which can be easily extended to work with masks too. This editor supports: * Loading movie files/image sequences * Build proxies with different size for loaded movie clip, also supports building undistorted proxies to increase speed of playback in undistorted mode. * Manual lens distortion mode calibration using grid and grease pencil * Supervised 2D tracking using two different algorithms KLT and SAD. * Basic algorithm for feature detection * Camera motion solving. scene orientation - New constraints to "link" scene objects with solved motions from clip: * Follow Track (make object follow 2D motion of track with given name or parent object to reconstructed 3D position of track) * Camera Solver to make camera moving in the same way as reconstructed camera This commit NOT includes changes from tomato branch: - New nodes (they'll be commited as separated patch) - Automatic image offset guessing for image input node and image editor (need to do more tests and gather more feedback) - Code cleanup in libmv-capi. It's not so critical cleanup, just increasing readability and understanadability of code. Better to make this chaneg when Keir will finish his current patch. More details about this project can be found on this page: http://wiki.blender.org/index.php/User:Nazg-gul/GSoC-2011 Further development of small features would be done in trunk, bigger/experimental features would first be implemented in tomato branch.
2011-11-07 12:55:18 +00:00
google::InitGoogleLogging(argv0);
google::SetCommandLineOption("logtostderr", "1");
google::SetCommandLineOption("v", "0");
google::SetCommandLineOption("stderrthreshold", severity_fatal);
google::SetCommandLineOption("minloglevel", severity_fatal);
Camera tracking integration =========================== Commiting camera tracking integration gsoc project into trunk. This commit includes: - Bundled version of libmv library (with some changes against official repo, re-sync with libmv repo a bit later) - New datatype ID called MovieClip which is optimized to work with movie clips (both of movie files and image sequences) and doing camera/motion tracking operations. - New editor called Clip Editor which is currently used for motion/tracking stuff only, but which can be easily extended to work with masks too. This editor supports: * Loading movie files/image sequences * Build proxies with different size for loaded movie clip, also supports building undistorted proxies to increase speed of playback in undistorted mode. * Manual lens distortion mode calibration using grid and grease pencil * Supervised 2D tracking using two different algorithms KLT and SAD. * Basic algorithm for feature detection * Camera motion solving. scene orientation - New constraints to "link" scene objects with solved motions from clip: * Follow Track (make object follow 2D motion of track with given name or parent object to reconstructed 3D position of track) * Camera Solver to make camera moving in the same way as reconstructed camera This commit NOT includes changes from tomato branch: - New nodes (they'll be commited as separated patch) - Automatic image offset guessing for image input node and image editor (need to do more tests and gather more feedback) - Code cleanup in libmv-capi. It's not so critical cleanup, just increasing readability and understanadability of code. Better to make this chaneg when Keir will finish his current patch. More details about this project can be found on this page: http://wiki.blender.org/index.php/User:Nazg-gul/GSoC-2011 Further development of small features would be done in trunk, bigger/experimental features would first be implemented in tomato branch.
2011-11-07 12:55:18 +00:00
}
void libmv_startDebugLogging(void)
{
google::SetCommandLineOption("logtostderr", "1");
Planar tracking support for motion tracking =========================================== Major list of changes done in tomato branch: - Add a planar tracking implementation to libmv This adds a new planar tracking implementation to libmv. The tracker is based on Ceres[1], the new nonlinear minimizer that myself and Sameer released from Google as open source. Since the motion model is more involved, the interface is different than the RegionTracker interface used previously in Blender. The start of a C API in libmv-capi.{cpp,h} is also included. - Migrate from pat_{min,max} for markers to 4 corners representation Convert markers in the movie clip editor / 2D tracker from using pat_min and pat_max notation to using the a more general, 4-corner representation. There is still considerable porting work to do; in particular sliding from preview widget does not work correct for rotated markers. All other areas should be ported to new representation: * Added support of sliding individual corners. LMB slide + Ctrl would scale the whole pattern * S would scale the whole marker, S-S would scale pattern only * Added support of marker's rotation which is currently rotates only patterns around their centers or all markers around median, Rotation or other non-translation/scaling transformation of search area doesn't make sense. * Track Preview widget would display transformed pattern which libmv actually operates with. - "Efficient Second-order Minimization" for the planar tracker This implements the "Efficient Second-order Minimization" scheme, as supported by the existing translation tracker. This increases the amount of per-iteration work, but decreases the number of iterations required to converge and also increases the size of the basin of attraction for the optimization. - Remove the use of the legacy RegionTracker API from Blender, and replaces it with the new TrackRegion API. This also adds several features to the planar tracker in libmv: * Do a brute-force initialization of tracking similar to "Hybrid" mode in the stable release, but using all floats. This is slower but more accurate. It is still necessary to evaluate if the performance loss is worth it. In particular, this change is necessary to support high bit depth imagery. * Add support for masks over the search window. This is a step towards supporting user-defined tracker masks. The tracker masks will make it easy for users to make a mask for e.g. a ball. Not exposed into interface yet/ * Add Pearson product moment correlation coefficient checking (aka "Correlation" in the UI. This causes tracking failure if the tracked patch is not linearly related to the template. * Add support for warping a few points in addition to the supplied points. This is useful because the tracking code deliberately does not expose the underlying warp representation. Instead, warps are specified in an aparametric way via the correspondences. - Replace the old style tracker configuration panel with the new planar tracking panel. From a users perspective, this means: * The old "tracking algorithm" picker is gone. There is only 1 algorithm now. We may revisit this later, but I would much prefer to have only 1 algorithm. So far no optimization work has been done so the speed is not there yet. * There is now a dropdown to select the motion model. Choices: * Translation * Translation, rotation * Translation, scale * Translation, rotation, scale * Affine * Perspective * The old "Hybrid" mode is gone; instead there is a toggle to enable or disable translation-only tracker initialization. This is the equivalent of the hyrbid mode before, but rewritten to work with the new planar tracking modes. * The pyramid levels setting is gone. At a future date, the planar tracker will decide to use pyramids or not automatically. The pyramid setting was ultimately a mistake; with the brute force initialization it is unnecessary. - Add light-normalized tracking Added the ability to normalize patterns by their average value while tracking, to make them invariant to global illumination changes. Additional details could be found at wiki page [2] [1] http://code.google.com/p/ceres-solver [2] http://wiki.blender.org/index.php/Dev:Ref/Release_Notes/2.64/Motion_Tracker
2012-06-10 15:28:19 +00:00
google::SetCommandLineOption("v", "2");
Camera tracking integration =========================== Commiting camera tracking integration gsoc project into trunk. This commit includes: - Bundled version of libmv library (with some changes against official repo, re-sync with libmv repo a bit later) - New datatype ID called MovieClip which is optimized to work with movie clips (both of movie files and image sequences) and doing camera/motion tracking operations. - New editor called Clip Editor which is currently used for motion/tracking stuff only, but which can be easily extended to work with masks too. This editor supports: * Loading movie files/image sequences * Build proxies with different size for loaded movie clip, also supports building undistorted proxies to increase speed of playback in undistorted mode. * Manual lens distortion mode calibration using grid and grease pencil * Supervised 2D tracking using two different algorithms KLT and SAD. * Basic algorithm for feature detection * Camera motion solving. scene orientation - New constraints to "link" scene objects with solved motions from clip: * Follow Track (make object follow 2D motion of track with given name or parent object to reconstructed 3D position of track) * Camera Solver to make camera moving in the same way as reconstructed camera This commit NOT includes changes from tomato branch: - New nodes (they'll be commited as separated patch) - Automatic image offset guessing for image input node and image editor (need to do more tests and gather more feedback) - Code cleanup in libmv-capi. It's not so critical cleanup, just increasing readability and understanadability of code. Better to make this chaneg when Keir will finish his current patch. More details about this project can be found on this page: http://wiki.blender.org/index.php/User:Nazg-gul/GSoC-2011 Further development of small features would be done in trunk, bigger/experimental features would first be implemented in tomato branch.
2011-11-07 12:55:18 +00:00
google::SetCommandLineOption("stderrthreshold", "1");
google::SetCommandLineOption("minloglevel", "0");
}
void libmv_setLoggingVerbosity(int verbosity)
{
char val[10];
snprintf(val, sizeof(val), "%d", verbosity);
google::SetCommandLineOption("v", val);
}
/* ************ Utility ************ */
Planar tracking support for motion tracking =========================================== Major list of changes done in tomato branch: - Add a planar tracking implementation to libmv This adds a new planar tracking implementation to libmv. The tracker is based on Ceres[1], the new nonlinear minimizer that myself and Sameer released from Google as open source. Since the motion model is more involved, the interface is different than the RegionTracker interface used previously in Blender. The start of a C API in libmv-capi.{cpp,h} is also included. - Migrate from pat_{min,max} for markers to 4 corners representation Convert markers in the movie clip editor / 2D tracker from using pat_min and pat_max notation to using the a more general, 4-corner representation. There is still considerable porting work to do; in particular sliding from preview widget does not work correct for rotated markers. All other areas should be ported to new representation: * Added support of sliding individual corners. LMB slide + Ctrl would scale the whole pattern * S would scale the whole marker, S-S would scale pattern only * Added support of marker's rotation which is currently rotates only patterns around their centers or all markers around median, Rotation or other non-translation/scaling transformation of search area doesn't make sense. * Track Preview widget would display transformed pattern which libmv actually operates with. - "Efficient Second-order Minimization" for the planar tracker This implements the "Efficient Second-order Minimization" scheme, as supported by the existing translation tracker. This increases the amount of per-iteration work, but decreases the number of iterations required to converge and also increases the size of the basin of attraction for the optimization. - Remove the use of the legacy RegionTracker API from Blender, and replaces it with the new TrackRegion API. This also adds several features to the planar tracker in libmv: * Do a brute-force initialization of tracking similar to "Hybrid" mode in the stable release, but using all floats. This is slower but more accurate. It is still necessary to evaluate if the performance loss is worth it. In particular, this change is necessary to support high bit depth imagery. * Add support for masks over the search window. This is a step towards supporting user-defined tracker masks. The tracker masks will make it easy for users to make a mask for e.g. a ball. Not exposed into interface yet/ * Add Pearson product moment correlation coefficient checking (aka "Correlation" in the UI. This causes tracking failure if the tracked patch is not linearly related to the template. * Add support for warping a few points in addition to the supplied points. This is useful because the tracking code deliberately does not expose the underlying warp representation. Instead, warps are specified in an aparametric way via the correspondences. - Replace the old style tracker configuration panel with the new planar tracking panel. From a users perspective, this means: * The old "tracking algorithm" picker is gone. There is only 1 algorithm now. We may revisit this later, but I would much prefer to have only 1 algorithm. So far no optimization work has been done so the speed is not there yet. * There is now a dropdown to select the motion model. Choices: * Translation * Translation, rotation * Translation, scale * Translation, rotation, scale * Affine * Perspective * The old "Hybrid" mode is gone; instead there is a toggle to enable or disable translation-only tracker initialization. This is the equivalent of the hyrbid mode before, but rewritten to work with the new planar tracking modes. * The pyramid levels setting is gone. At a future date, the planar tracker will decide to use pyramids or not automatically. The pyramid setting was ultimately a mistake; with the brute force initialization it is unnecessary. - Add light-normalized tracking Added the ability to normalize patterns by their average value while tracking, to make them invariant to global illumination changes. Additional details could be found at wiki page [2] [1] http://code.google.com/p/ceres-solver [2] http://wiki.blender.org/index.php/Dev:Ref/Release_Notes/2.64/Motion_Tracker
2012-06-10 15:28:19 +00:00
static void floatBufToImage(const float *buf, int width, int height, int channels, libmv::FloatImage *image)
Camera tracking integration =========================== Commiting camera tracking integration gsoc project into trunk. This commit includes: - Bundled version of libmv library (with some changes against official repo, re-sync with libmv repo a bit later) - New datatype ID called MovieClip which is optimized to work with movie clips (both of movie files and image sequences) and doing camera/motion tracking operations. - New editor called Clip Editor which is currently used for motion/tracking stuff only, but which can be easily extended to work with masks too. This editor supports: * Loading movie files/image sequences * Build proxies with different size for loaded movie clip, also supports building undistorted proxies to increase speed of playback in undistorted mode. * Manual lens distortion mode calibration using grid and grease pencil * Supervised 2D tracking using two different algorithms KLT and SAD. * Basic algorithm for feature detection * Camera motion solving. scene orientation - New constraints to "link" scene objects with solved motions from clip: * Follow Track (make object follow 2D motion of track with given name or parent object to reconstructed 3D position of track) * Camera Solver to make camera moving in the same way as reconstructed camera This commit NOT includes changes from tomato branch: - New nodes (they'll be commited as separated patch) - Automatic image offset guessing for image input node and image editor (need to do more tests and gather more feedback) - Code cleanup in libmv-capi. It's not so critical cleanup, just increasing readability and understanadability of code. Better to make this chaneg when Keir will finish his current patch. More details about this project can be found on this page: http://wiki.blender.org/index.php/User:Nazg-gul/GSoC-2011 Further development of small features would be done in trunk, bigger/experimental features would first be implemented in tomato branch.
2011-11-07 12:55:18 +00:00
{
Planar tracking support for motion tracking =========================================== Major list of changes done in tomato branch: - Add a planar tracking implementation to libmv This adds a new planar tracking implementation to libmv. The tracker is based on Ceres[1], the new nonlinear minimizer that myself and Sameer released from Google as open source. Since the motion model is more involved, the interface is different than the RegionTracker interface used previously in Blender. The start of a C API in libmv-capi.{cpp,h} is also included. - Migrate from pat_{min,max} for markers to 4 corners representation Convert markers in the movie clip editor / 2D tracker from using pat_min and pat_max notation to using the a more general, 4-corner representation. There is still considerable porting work to do; in particular sliding from preview widget does not work correct for rotated markers. All other areas should be ported to new representation: * Added support of sliding individual corners. LMB slide + Ctrl would scale the whole pattern * S would scale the whole marker, S-S would scale pattern only * Added support of marker's rotation which is currently rotates only patterns around their centers or all markers around median, Rotation or other non-translation/scaling transformation of search area doesn't make sense. * Track Preview widget would display transformed pattern which libmv actually operates with. - "Efficient Second-order Minimization" for the planar tracker This implements the "Efficient Second-order Minimization" scheme, as supported by the existing translation tracker. This increases the amount of per-iteration work, but decreases the number of iterations required to converge and also increases the size of the basin of attraction for the optimization. - Remove the use of the legacy RegionTracker API from Blender, and replaces it with the new TrackRegion API. This also adds several features to the planar tracker in libmv: * Do a brute-force initialization of tracking similar to "Hybrid" mode in the stable release, but using all floats. This is slower but more accurate. It is still necessary to evaluate if the performance loss is worth it. In particular, this change is necessary to support high bit depth imagery. * Add support for masks over the search window. This is a step towards supporting user-defined tracker masks. The tracker masks will make it easy for users to make a mask for e.g. a ball. Not exposed into interface yet/ * Add Pearson product moment correlation coefficient checking (aka "Correlation" in the UI. This causes tracking failure if the tracked patch is not linearly related to the template. * Add support for warping a few points in addition to the supplied points. This is useful because the tracking code deliberately does not expose the underlying warp representation. Instead, warps are specified in an aparametric way via the correspondences. - Replace the old style tracker configuration panel with the new planar tracking panel. From a users perspective, this means: * The old "tracking algorithm" picker is gone. There is only 1 algorithm now. We may revisit this later, but I would much prefer to have only 1 algorithm. So far no optimization work has been done so the speed is not there yet. * There is now a dropdown to select the motion model. Choices: * Translation * Translation, rotation * Translation, scale * Translation, rotation, scale * Affine * Perspective * The old "Hybrid" mode is gone; instead there is a toggle to enable or disable translation-only tracker initialization. This is the equivalent of the hyrbid mode before, but rewritten to work with the new planar tracking modes. * The pyramid levels setting is gone. At a future date, the planar tracker will decide to use pyramids or not automatically. The pyramid setting was ultimately a mistake; with the brute force initialization it is unnecessary. - Add light-normalized tracking Added the ability to normalize patterns by their average value while tracking, to make them invariant to global illumination changes. Additional details could be found at wiki page [2] [1] http://code.google.com/p/ceres-solver [2] http://wiki.blender.org/index.php/Dev:Ref/Release_Notes/2.64/Motion_Tracker
2012-06-10 15:28:19 +00:00
int x, y, k, a = 0;
Camera tracking integration =========================== Commiting camera tracking integration gsoc project into trunk. This commit includes: - Bundled version of libmv library (with some changes against official repo, re-sync with libmv repo a bit later) - New datatype ID called MovieClip which is optimized to work with movie clips (both of movie files and image sequences) and doing camera/motion tracking operations. - New editor called Clip Editor which is currently used for motion/tracking stuff only, but which can be easily extended to work with masks too. This editor supports: * Loading movie files/image sequences * Build proxies with different size for loaded movie clip, also supports building undistorted proxies to increase speed of playback in undistorted mode. * Manual lens distortion mode calibration using grid and grease pencil * Supervised 2D tracking using two different algorithms KLT and SAD. * Basic algorithm for feature detection * Camera motion solving. scene orientation - New constraints to "link" scene objects with solved motions from clip: * Follow Track (make object follow 2D motion of track with given name or parent object to reconstructed 3D position of track) * Camera Solver to make camera moving in the same way as reconstructed camera This commit NOT includes changes from tomato branch: - New nodes (they'll be commited as separated patch) - Automatic image offset guessing for image input node and image editor (need to do more tests and gather more feedback) - Code cleanup in libmv-capi. It's not so critical cleanup, just increasing readability and understanadability of code. Better to make this chaneg when Keir will finish his current patch. More details about this project can be found on this page: http://wiki.blender.org/index.php/User:Nazg-gul/GSoC-2011 Further development of small features would be done in trunk, bigger/experimental features would first be implemented in tomato branch.
2011-11-07 12:55:18 +00:00
Planar tracking support for motion tracking =========================================== Major list of changes done in tomato branch: - Add a planar tracking implementation to libmv This adds a new planar tracking implementation to libmv. The tracker is based on Ceres[1], the new nonlinear minimizer that myself and Sameer released from Google as open source. Since the motion model is more involved, the interface is different than the RegionTracker interface used previously in Blender. The start of a C API in libmv-capi.{cpp,h} is also included. - Migrate from pat_{min,max} for markers to 4 corners representation Convert markers in the movie clip editor / 2D tracker from using pat_min and pat_max notation to using the a more general, 4-corner representation. There is still considerable porting work to do; in particular sliding from preview widget does not work correct for rotated markers. All other areas should be ported to new representation: * Added support of sliding individual corners. LMB slide + Ctrl would scale the whole pattern * S would scale the whole marker, S-S would scale pattern only * Added support of marker's rotation which is currently rotates only patterns around their centers or all markers around median, Rotation or other non-translation/scaling transformation of search area doesn't make sense. * Track Preview widget would display transformed pattern which libmv actually operates with. - "Efficient Second-order Minimization" for the planar tracker This implements the "Efficient Second-order Minimization" scheme, as supported by the existing translation tracker. This increases the amount of per-iteration work, but decreases the number of iterations required to converge and also increases the size of the basin of attraction for the optimization. - Remove the use of the legacy RegionTracker API from Blender, and replaces it with the new TrackRegion API. This also adds several features to the planar tracker in libmv: * Do a brute-force initialization of tracking similar to "Hybrid" mode in the stable release, but using all floats. This is slower but more accurate. It is still necessary to evaluate if the performance loss is worth it. In particular, this change is necessary to support high bit depth imagery. * Add support for masks over the search window. This is a step towards supporting user-defined tracker masks. The tracker masks will make it easy for users to make a mask for e.g. a ball. Not exposed into interface yet/ * Add Pearson product moment correlation coefficient checking (aka "Correlation" in the UI. This causes tracking failure if the tracked patch is not linearly related to the template. * Add support for warping a few points in addition to the supplied points. This is useful because the tracking code deliberately does not expose the underlying warp representation. Instead, warps are specified in an aparametric way via the correspondences. - Replace the old style tracker configuration panel with the new planar tracking panel. From a users perspective, this means: * The old "tracking algorithm" picker is gone. There is only 1 algorithm now. We may revisit this later, but I would much prefer to have only 1 algorithm. So far no optimization work has been done so the speed is not there yet. * There is now a dropdown to select the motion model. Choices: * Translation * Translation, rotation * Translation, scale * Translation, rotation, scale * Affine * Perspective * The old "Hybrid" mode is gone; instead there is a toggle to enable or disable translation-only tracker initialization. This is the equivalent of the hyrbid mode before, but rewritten to work with the new planar tracking modes. * The pyramid levels setting is gone. At a future date, the planar tracker will decide to use pyramids or not automatically. The pyramid setting was ultimately a mistake; with the brute force initialization it is unnecessary. - Add light-normalized tracking Added the ability to normalize patterns by their average value while tracking, to make them invariant to global illumination changes. Additional details could be found at wiki page [2] [1] http://code.google.com/p/ceres-solver [2] http://wiki.blender.org/index.php/Dev:Ref/Release_Notes/2.64/Motion_Tracker
2012-06-10 15:28:19 +00:00
image->Resize(height, width, channels);
Camera tracking integration =========================== Commiting camera tracking integration gsoc project into trunk. This commit includes: - Bundled version of libmv library (with some changes against official repo, re-sync with libmv repo a bit later) - New datatype ID called MovieClip which is optimized to work with movie clips (both of movie files and image sequences) and doing camera/motion tracking operations. - New editor called Clip Editor which is currently used for motion/tracking stuff only, but which can be easily extended to work with masks too. This editor supports: * Loading movie files/image sequences * Build proxies with different size for loaded movie clip, also supports building undistorted proxies to increase speed of playback in undistorted mode. * Manual lens distortion mode calibration using grid and grease pencil * Supervised 2D tracking using two different algorithms KLT and SAD. * Basic algorithm for feature detection * Camera motion solving. scene orientation - New constraints to "link" scene objects with solved motions from clip: * Follow Track (make object follow 2D motion of track with given name or parent object to reconstructed 3D position of track) * Camera Solver to make camera moving in the same way as reconstructed camera This commit NOT includes changes from tomato branch: - New nodes (they'll be commited as separated patch) - Automatic image offset guessing for image input node and image editor (need to do more tests and gather more feedback) - Code cleanup in libmv-capi. It's not so critical cleanup, just increasing readability and understanadability of code. Better to make this chaneg when Keir will finish his current patch. More details about this project can be found on this page: http://wiki.blender.org/index.php/User:Nazg-gul/GSoC-2011 Further development of small features would be done in trunk, bigger/experimental features would first be implemented in tomato branch.
2011-11-07 12:55:18 +00:00
for (y = 0; y < height; y++) {
for (x = 0; x < width; x++) {
Planar tracking support for motion tracking =========================================== Major list of changes done in tomato branch: - Add a planar tracking implementation to libmv This adds a new planar tracking implementation to libmv. The tracker is based on Ceres[1], the new nonlinear minimizer that myself and Sameer released from Google as open source. Since the motion model is more involved, the interface is different than the RegionTracker interface used previously in Blender. The start of a C API in libmv-capi.{cpp,h} is also included. - Migrate from pat_{min,max} for markers to 4 corners representation Convert markers in the movie clip editor / 2D tracker from using pat_min and pat_max notation to using the a more general, 4-corner representation. There is still considerable porting work to do; in particular sliding from preview widget does not work correct for rotated markers. All other areas should be ported to new representation: * Added support of sliding individual corners. LMB slide + Ctrl would scale the whole pattern * S would scale the whole marker, S-S would scale pattern only * Added support of marker's rotation which is currently rotates only patterns around their centers or all markers around median, Rotation or other non-translation/scaling transformation of search area doesn't make sense. * Track Preview widget would display transformed pattern which libmv actually operates with. - "Efficient Second-order Minimization" for the planar tracker This implements the "Efficient Second-order Minimization" scheme, as supported by the existing translation tracker. This increases the amount of per-iteration work, but decreases the number of iterations required to converge and also increases the size of the basin of attraction for the optimization. - Remove the use of the legacy RegionTracker API from Blender, and replaces it with the new TrackRegion API. This also adds several features to the planar tracker in libmv: * Do a brute-force initialization of tracking similar to "Hybrid" mode in the stable release, but using all floats. This is slower but more accurate. It is still necessary to evaluate if the performance loss is worth it. In particular, this change is necessary to support high bit depth imagery. * Add support for masks over the search window. This is a step towards supporting user-defined tracker masks. The tracker masks will make it easy for users to make a mask for e.g. a ball. Not exposed into interface yet/ * Add Pearson product moment correlation coefficient checking (aka "Correlation" in the UI. This causes tracking failure if the tracked patch is not linearly related to the template. * Add support for warping a few points in addition to the supplied points. This is useful because the tracking code deliberately does not expose the underlying warp representation. Instead, warps are specified in an aparametric way via the correspondences. - Replace the old style tracker configuration panel with the new planar tracking panel. From a users perspective, this means: * The old "tracking algorithm" picker is gone. There is only 1 algorithm now. We may revisit this later, but I would much prefer to have only 1 algorithm. So far no optimization work has been done so the speed is not there yet. * There is now a dropdown to select the motion model. Choices: * Translation * Translation, rotation * Translation, scale * Translation, rotation, scale * Affine * Perspective * The old "Hybrid" mode is gone; instead there is a toggle to enable or disable translation-only tracker initialization. This is the equivalent of the hyrbid mode before, but rewritten to work with the new planar tracking modes. * The pyramid levels setting is gone. At a future date, the planar tracker will decide to use pyramids or not automatically. The pyramid setting was ultimately a mistake; with the brute force initialization it is unnecessary. - Add light-normalized tracking Added the ability to normalize patterns by their average value while tracking, to make them invariant to global illumination changes. Additional details could be found at wiki page [2] [1] http://code.google.com/p/ceres-solver [2] http://wiki.blender.org/index.php/Dev:Ref/Release_Notes/2.64/Motion_Tracker
2012-06-10 15:28:19 +00:00
for (k = 0; k < channels; k++) {
(*image)(y, x, k) = buf[a++];
}
}
}
}
static void imageToFloatBuf(const libmv::FloatImage *image, int channels, float *buf)
{
int x, y, k, a = 0;
for (y = 0; y < image->Height(); y++) {
for (x = 0; x < image->Width(); x++) {
for (k = 0; k < channels; k++) {
buf[a++] = (*image)(y, x, k);
}
Camera tracking integration =========================== Commiting camera tracking integration gsoc project into trunk. This commit includes: - Bundled version of libmv library (with some changes against official repo, re-sync with libmv repo a bit later) - New datatype ID called MovieClip which is optimized to work with movie clips (both of movie files and image sequences) and doing camera/motion tracking operations. - New editor called Clip Editor which is currently used for motion/tracking stuff only, but which can be easily extended to work with masks too. This editor supports: * Loading movie files/image sequences * Build proxies with different size for loaded movie clip, also supports building undistorted proxies to increase speed of playback in undistorted mode. * Manual lens distortion mode calibration using grid and grease pencil * Supervised 2D tracking using two different algorithms KLT and SAD. * Basic algorithm for feature detection * Camera motion solving. scene orientation - New constraints to "link" scene objects with solved motions from clip: * Follow Track (make object follow 2D motion of track with given name or parent object to reconstructed 3D position of track) * Camera Solver to make camera moving in the same way as reconstructed camera This commit NOT includes changes from tomato branch: - New nodes (they'll be commited as separated patch) - Automatic image offset guessing for image input node and image editor (need to do more tests and gather more feedback) - Code cleanup in libmv-capi. It's not so critical cleanup, just increasing readability and understanadability of code. Better to make this chaneg when Keir will finish his current patch. More details about this project can be found on this page: http://wiki.blender.org/index.php/User:Nazg-gul/GSoC-2011 Further development of small features would be done in trunk, bigger/experimental features would first be implemented in tomato branch.
2011-11-07 12:55:18 +00:00
}
}
}
Planar tracking support for motion tracking =========================================== Major list of changes done in tomato branch: - Add a planar tracking implementation to libmv This adds a new planar tracking implementation to libmv. The tracker is based on Ceres[1], the new nonlinear minimizer that myself and Sameer released from Google as open source. Since the motion model is more involved, the interface is different than the RegionTracker interface used previously in Blender. The start of a C API in libmv-capi.{cpp,h} is also included. - Migrate from pat_{min,max} for markers to 4 corners representation Convert markers in the movie clip editor / 2D tracker from using pat_min and pat_max notation to using the a more general, 4-corner representation. There is still considerable porting work to do; in particular sliding from preview widget does not work correct for rotated markers. All other areas should be ported to new representation: * Added support of sliding individual corners. LMB slide + Ctrl would scale the whole pattern * S would scale the whole marker, S-S would scale pattern only * Added support of marker's rotation which is currently rotates only patterns around their centers or all markers around median, Rotation or other non-translation/scaling transformation of search area doesn't make sense. * Track Preview widget would display transformed pattern which libmv actually operates with. - "Efficient Second-order Minimization" for the planar tracker This implements the "Efficient Second-order Minimization" scheme, as supported by the existing translation tracker. This increases the amount of per-iteration work, but decreases the number of iterations required to converge and also increases the size of the basin of attraction for the optimization. - Remove the use of the legacy RegionTracker API from Blender, and replaces it with the new TrackRegion API. This also adds several features to the planar tracker in libmv: * Do a brute-force initialization of tracking similar to "Hybrid" mode in the stable release, but using all floats. This is slower but more accurate. It is still necessary to evaluate if the performance loss is worth it. In particular, this change is necessary to support high bit depth imagery. * Add support for masks over the search window. This is a step towards supporting user-defined tracker masks. The tracker masks will make it easy for users to make a mask for e.g. a ball. Not exposed into interface yet/ * Add Pearson product moment correlation coefficient checking (aka "Correlation" in the UI. This causes tracking failure if the tracked patch is not linearly related to the template. * Add support for warping a few points in addition to the supplied points. This is useful because the tracking code deliberately does not expose the underlying warp representation. Instead, warps are specified in an aparametric way via the correspondences. - Replace the old style tracker configuration panel with the new planar tracking panel. From a users perspective, this means: * The old "tracking algorithm" picker is gone. There is only 1 algorithm now. We may revisit this later, but I would much prefer to have only 1 algorithm. So far no optimization work has been done so the speed is not there yet. * There is now a dropdown to select the motion model. Choices: * Translation * Translation, rotation * Translation, scale * Translation, rotation, scale * Affine * Perspective * The old "Hybrid" mode is gone; instead there is a toggle to enable or disable translation-only tracker initialization. This is the equivalent of the hyrbid mode before, but rewritten to work with the new planar tracking modes. * The pyramid levels setting is gone. At a future date, the planar tracker will decide to use pyramids or not automatically. The pyramid setting was ultimately a mistake; with the brute force initialization it is unnecessary. - Add light-normalized tracking Added the ability to normalize patterns by their average value while tracking, to make them invariant to global illumination changes. Additional details could be found at wiki page [2] [1] http://code.google.com/p/ceres-solver [2] http://wiki.blender.org/index.php/Dev:Ref/Release_Notes/2.64/Motion_Tracker
2012-06-10 15:28:19 +00:00
#if defined(DUMP_FAILURE) || defined (DUMP_ALWAYS)
static void savePNGImage(png_bytep *row_pointers, int width, int height, int depth, int color_type, char *file_name)
Camera tracking integration =========================== Commiting camera tracking integration gsoc project into trunk. This commit includes: - Bundled version of libmv library (with some changes against official repo, re-sync with libmv repo a bit later) - New datatype ID called MovieClip which is optimized to work with movie clips (both of movie files and image sequences) and doing camera/motion tracking operations. - New editor called Clip Editor which is currently used for motion/tracking stuff only, but which can be easily extended to work with masks too. This editor supports: * Loading movie files/image sequences * Build proxies with different size for loaded movie clip, also supports building undistorted proxies to increase speed of playback in undistorted mode. * Manual lens distortion mode calibration using grid and grease pencil * Supervised 2D tracking using two different algorithms KLT and SAD. * Basic algorithm for feature detection * Camera motion solving. scene orientation - New constraints to "link" scene objects with solved motions from clip: * Follow Track (make object follow 2D motion of track with given name or parent object to reconstructed 3D position of track) * Camera Solver to make camera moving in the same way as reconstructed camera This commit NOT includes changes from tomato branch: - New nodes (they'll be commited as separated patch) - Automatic image offset guessing for image input node and image editor (need to do more tests and gather more feedback) - Code cleanup in libmv-capi. It's not so critical cleanup, just increasing readability and understanadability of code. Better to make this chaneg when Keir will finish his current patch. More details about this project can be found on this page: http://wiki.blender.org/index.php/User:Nazg-gul/GSoC-2011 Further development of small features would be done in trunk, bigger/experimental features would first be implemented in tomato branch.
2011-11-07 12:55:18 +00:00
{
png_infop info_ptr;
png_structp png_ptr;
FILE *fp = fopen(file_name, "wb");
if (!fp)
return;
/* Initialize stuff */
png_ptr = png_create_write_struct(PNG_LIBPNG_VER_STRING, NULL, NULL, NULL);
info_ptr = png_create_info_struct(png_ptr);
if (setjmp(png_jmpbuf(png_ptr))) {
fclose(fp);
return;
}
png_init_io(png_ptr, fp);
/* write header */
if (setjmp(png_jmpbuf(png_ptr))) {
fclose(fp);
return;
}
png_set_IHDR(png_ptr, info_ptr, width, height,
depth, color_type, PNG_INTERLACE_NONE,
PNG_COMPRESSION_TYPE_BASE, PNG_FILTER_TYPE_BASE);
png_write_info(png_ptr, info_ptr);
/* write bytes */
if (setjmp(png_jmpbuf(png_ptr))) {
fclose(fp);
return;
}
png_write_image(png_ptr, row_pointers);
/* end write */
if (setjmp(png_jmpbuf(png_ptr))) {
fclose(fp);
return;
}
png_write_end(png_ptr, NULL);
fclose(fp);
}
static void saveImage(const char *prefix, libmv::FloatImage image, int x0, int y0)
Camera tracking integration =========================== Commiting camera tracking integration gsoc project into trunk. This commit includes: - Bundled version of libmv library (with some changes against official repo, re-sync with libmv repo a bit later) - New datatype ID called MovieClip which is optimized to work with movie clips (both of movie files and image sequences) and doing camera/motion tracking operations. - New editor called Clip Editor which is currently used for motion/tracking stuff only, but which can be easily extended to work with masks too. This editor supports: * Loading movie files/image sequences * Build proxies with different size for loaded movie clip, also supports building undistorted proxies to increase speed of playback in undistorted mode. * Manual lens distortion mode calibration using grid and grease pencil * Supervised 2D tracking using two different algorithms KLT and SAD. * Basic algorithm for feature detection * Camera motion solving. scene orientation - New constraints to "link" scene objects with solved motions from clip: * Follow Track (make object follow 2D motion of track with given name or parent object to reconstructed 3D position of track) * Camera Solver to make camera moving in the same way as reconstructed camera This commit NOT includes changes from tomato branch: - New nodes (they'll be commited as separated patch) - Automatic image offset guessing for image input node and image editor (need to do more tests and gather more feedback) - Code cleanup in libmv-capi. It's not so critical cleanup, just increasing readability and understanadability of code. Better to make this chaneg when Keir will finish his current patch. More details about this project can be found on this page: http://wiki.blender.org/index.php/User:Nazg-gul/GSoC-2011 Further development of small features would be done in trunk, bigger/experimental features would first be implemented in tomato branch.
2011-11-07 12:55:18 +00:00
{
int x, y;
png_bytep *row_pointers;
row_pointers= (png_bytep*)malloc(sizeof(png_bytep)*image.Height());
for (y = 0; y < image.Height(); y++) {
row_pointers[y]= (png_bytep)malloc(sizeof(png_byte)*4*image.Width());
for (x = 0; x < image.Width(); x++) {
Planar tracking support for motion tracking =========================================== Major list of changes done in tomato branch: - Add a planar tracking implementation to libmv This adds a new planar tracking implementation to libmv. The tracker is based on Ceres[1], the new nonlinear minimizer that myself and Sameer released from Google as open source. Since the motion model is more involved, the interface is different than the RegionTracker interface used previously in Blender. The start of a C API in libmv-capi.{cpp,h} is also included. - Migrate from pat_{min,max} for markers to 4 corners representation Convert markers in the movie clip editor / 2D tracker from using pat_min and pat_max notation to using the a more general, 4-corner representation. There is still considerable porting work to do; in particular sliding from preview widget does not work correct for rotated markers. All other areas should be ported to new representation: * Added support of sliding individual corners. LMB slide + Ctrl would scale the whole pattern * S would scale the whole marker, S-S would scale pattern only * Added support of marker's rotation which is currently rotates only patterns around their centers or all markers around median, Rotation or other non-translation/scaling transformation of search area doesn't make sense. * Track Preview widget would display transformed pattern which libmv actually operates with. - "Efficient Second-order Minimization" for the planar tracker This implements the "Efficient Second-order Minimization" scheme, as supported by the existing translation tracker. This increases the amount of per-iteration work, but decreases the number of iterations required to converge and also increases the size of the basin of attraction for the optimization. - Remove the use of the legacy RegionTracker API from Blender, and replaces it with the new TrackRegion API. This also adds several features to the planar tracker in libmv: * Do a brute-force initialization of tracking similar to "Hybrid" mode in the stable release, but using all floats. This is slower but more accurate. It is still necessary to evaluate if the performance loss is worth it. In particular, this change is necessary to support high bit depth imagery. * Add support for masks over the search window. This is a step towards supporting user-defined tracker masks. The tracker masks will make it easy for users to make a mask for e.g. a ball. Not exposed into interface yet/ * Add Pearson product moment correlation coefficient checking (aka "Correlation" in the UI. This causes tracking failure if the tracked patch is not linearly related to the template. * Add support for warping a few points in addition to the supplied points. This is useful because the tracking code deliberately does not expose the underlying warp representation. Instead, warps are specified in an aparametric way via the correspondences. - Replace the old style tracker configuration panel with the new planar tracking panel. From a users perspective, this means: * The old "tracking algorithm" picker is gone. There is only 1 algorithm now. We may revisit this later, but I would much prefer to have only 1 algorithm. So far no optimization work has been done so the speed is not there yet. * There is now a dropdown to select the motion model. Choices: * Translation * Translation, rotation * Translation, scale * Translation, rotation, scale * Affine * Perspective * The old "Hybrid" mode is gone; instead there is a toggle to enable or disable translation-only tracker initialization. This is the equivalent of the hyrbid mode before, but rewritten to work with the new planar tracking modes. * The pyramid levels setting is gone. At a future date, the planar tracker will decide to use pyramids or not automatically. The pyramid setting was ultimately a mistake; with the brute force initialization it is unnecessary. - Add light-normalized tracking Added the ability to normalize patterns by their average value while tracking, to make them invariant to global illumination changes. Additional details could be found at wiki page [2] [1] http://code.google.com/p/ceres-solver [2] http://wiki.blender.org/index.php/Dev:Ref/Release_Notes/2.64/Motion_Tracker
2012-06-10 15:28:19 +00:00
if (x0 == x && image.Height() - y0 - 1 == y) {
Camera tracking integration =========================== Commiting camera tracking integration gsoc project into trunk. This commit includes: - Bundled version of libmv library (with some changes against official repo, re-sync with libmv repo a bit later) - New datatype ID called MovieClip which is optimized to work with movie clips (both of movie files and image sequences) and doing camera/motion tracking operations. - New editor called Clip Editor which is currently used for motion/tracking stuff only, but which can be easily extended to work with masks too. This editor supports: * Loading movie files/image sequences * Build proxies with different size for loaded movie clip, also supports building undistorted proxies to increase speed of playback in undistorted mode. * Manual lens distortion mode calibration using grid and grease pencil * Supervised 2D tracking using two different algorithms KLT and SAD. * Basic algorithm for feature detection * Camera motion solving. scene orientation - New constraints to "link" scene objects with solved motions from clip: * Follow Track (make object follow 2D motion of track with given name or parent object to reconstructed 3D position of track) * Camera Solver to make camera moving in the same way as reconstructed camera This commit NOT includes changes from tomato branch: - New nodes (they'll be commited as separated patch) - Automatic image offset guessing for image input node and image editor (need to do more tests and gather more feedback) - Code cleanup in libmv-capi. It's not so critical cleanup, just increasing readability and understanadability of code. Better to make this chaneg when Keir will finish his current patch. More details about this project can be found on this page: http://wiki.blender.org/index.php/User:Nazg-gul/GSoC-2011 Further development of small features would be done in trunk, bigger/experimental features would first be implemented in tomato branch.
2011-11-07 12:55:18 +00:00
row_pointers[y][x*4+0]= 255;
row_pointers[y][x*4+1]= 0;
row_pointers[y][x*4+2]= 0;
row_pointers[y][x*4+3]= 255;
}
else {
Planar tracking support for motion tracking =========================================== Major list of changes done in tomato branch: - Add a planar tracking implementation to libmv This adds a new planar tracking implementation to libmv. The tracker is based on Ceres[1], the new nonlinear minimizer that myself and Sameer released from Google as open source. Since the motion model is more involved, the interface is different than the RegionTracker interface used previously in Blender. The start of a C API in libmv-capi.{cpp,h} is also included. - Migrate from pat_{min,max} for markers to 4 corners representation Convert markers in the movie clip editor / 2D tracker from using pat_min and pat_max notation to using the a more general, 4-corner representation. There is still considerable porting work to do; in particular sliding from preview widget does not work correct for rotated markers. All other areas should be ported to new representation: * Added support of sliding individual corners. LMB slide + Ctrl would scale the whole pattern * S would scale the whole marker, S-S would scale pattern only * Added support of marker's rotation which is currently rotates only patterns around their centers or all markers around median, Rotation or other non-translation/scaling transformation of search area doesn't make sense. * Track Preview widget would display transformed pattern which libmv actually operates with. - "Efficient Second-order Minimization" for the planar tracker This implements the "Efficient Second-order Minimization" scheme, as supported by the existing translation tracker. This increases the amount of per-iteration work, but decreases the number of iterations required to converge and also increases the size of the basin of attraction for the optimization. - Remove the use of the legacy RegionTracker API from Blender, and replaces it with the new TrackRegion API. This also adds several features to the planar tracker in libmv: * Do a brute-force initialization of tracking similar to "Hybrid" mode in the stable release, but using all floats. This is slower but more accurate. It is still necessary to evaluate if the performance loss is worth it. In particular, this change is necessary to support high bit depth imagery. * Add support for masks over the search window. This is a step towards supporting user-defined tracker masks. The tracker masks will make it easy for users to make a mask for e.g. a ball. Not exposed into interface yet/ * Add Pearson product moment correlation coefficient checking (aka "Correlation" in the UI. This causes tracking failure if the tracked patch is not linearly related to the template. * Add support for warping a few points in addition to the supplied points. This is useful because the tracking code deliberately does not expose the underlying warp representation. Instead, warps are specified in an aparametric way via the correspondences. - Replace the old style tracker configuration panel with the new planar tracking panel. From a users perspective, this means: * The old "tracking algorithm" picker is gone. There is only 1 algorithm now. We may revisit this later, but I would much prefer to have only 1 algorithm. So far no optimization work has been done so the speed is not there yet. * There is now a dropdown to select the motion model. Choices: * Translation * Translation, rotation * Translation, scale * Translation, rotation, scale * Affine * Perspective * The old "Hybrid" mode is gone; instead there is a toggle to enable or disable translation-only tracker initialization. This is the equivalent of the hyrbid mode before, but rewritten to work with the new planar tracking modes. * The pyramid levels setting is gone. At a future date, the planar tracker will decide to use pyramids or not automatically. The pyramid setting was ultimately a mistake; with the brute force initialization it is unnecessary. - Add light-normalized tracking Added the ability to normalize patterns by their average value while tracking, to make them invariant to global illumination changes. Additional details could be found at wiki page [2] [1] http://code.google.com/p/ceres-solver [2] http://wiki.blender.org/index.php/Dev:Ref/Release_Notes/2.64/Motion_Tracker
2012-06-10 15:28:19 +00:00
float pixel = image(image.Height() - y - 1, x, 0);
Camera tracking integration =========================== Commiting camera tracking integration gsoc project into trunk. This commit includes: - Bundled version of libmv library (with some changes against official repo, re-sync with libmv repo a bit later) - New datatype ID called MovieClip which is optimized to work with movie clips (both of movie files and image sequences) and doing camera/motion tracking operations. - New editor called Clip Editor which is currently used for motion/tracking stuff only, but which can be easily extended to work with masks too. This editor supports: * Loading movie files/image sequences * Build proxies with different size for loaded movie clip, also supports building undistorted proxies to increase speed of playback in undistorted mode. * Manual lens distortion mode calibration using grid and grease pencil * Supervised 2D tracking using two different algorithms KLT and SAD. * Basic algorithm for feature detection * Camera motion solving. scene orientation - New constraints to "link" scene objects with solved motions from clip: * Follow Track (make object follow 2D motion of track with given name or parent object to reconstructed 3D position of track) * Camera Solver to make camera moving in the same way as reconstructed camera This commit NOT includes changes from tomato branch: - New nodes (they'll be commited as separated patch) - Automatic image offset guessing for image input node and image editor (need to do more tests and gather more feedback) - Code cleanup in libmv-capi. It's not so critical cleanup, just increasing readability and understanadability of code. Better to make this chaneg when Keir will finish his current patch. More details about this project can be found on this page: http://wiki.blender.org/index.php/User:Nazg-gul/GSoC-2011 Further development of small features would be done in trunk, bigger/experimental features would first be implemented in tomato branch.
2011-11-07 12:55:18 +00:00
row_pointers[y][x*4+0]= pixel*255;
row_pointers[y][x*4+1]= pixel*255;
row_pointers[y][x*4+2]= pixel*255;
row_pointers[y][x*4+3]= 255;
}
}
}
{
static int a= 0;
char buf[128];
snprintf(buf, sizeof(buf), "%s_%02d.png", prefix, ++a);
savePNGImage(row_pointers, image.Width(), image.Height(), 8, PNG_COLOR_TYPE_RGBA, buf);
}
for (y = 0; y < image.Height(); y++) {
free(row_pointers[y]);
}
free(row_pointers);
}
static void saveBytesImage(const char *prefix, unsigned char *data, int width, int height)
Camera tracking integration =========================== Commiting camera tracking integration gsoc project into trunk. This commit includes: - Bundled version of libmv library (with some changes against official repo, re-sync with libmv repo a bit later) - New datatype ID called MovieClip which is optimized to work with movie clips (both of movie files and image sequences) and doing camera/motion tracking operations. - New editor called Clip Editor which is currently used for motion/tracking stuff only, but which can be easily extended to work with masks too. This editor supports: * Loading movie files/image sequences * Build proxies with different size for loaded movie clip, also supports building undistorted proxies to increase speed of playback in undistorted mode. * Manual lens distortion mode calibration using grid and grease pencil * Supervised 2D tracking using two different algorithms KLT and SAD. * Basic algorithm for feature detection * Camera motion solving. scene orientation - New constraints to "link" scene objects with solved motions from clip: * Follow Track (make object follow 2D motion of track with given name or parent object to reconstructed 3D position of track) * Camera Solver to make camera moving in the same way as reconstructed camera This commit NOT includes changes from tomato branch: - New nodes (they'll be commited as separated patch) - Automatic image offset guessing for image input node and image editor (need to do more tests and gather more feedback) - Code cleanup in libmv-capi. It's not so critical cleanup, just increasing readability and understanadability of code. Better to make this chaneg when Keir will finish his current patch. More details about this project can be found on this page: http://wiki.blender.org/index.php/User:Nazg-gul/GSoC-2011 Further development of small features would be done in trunk, bigger/experimental features would first be implemented in tomato branch.
2011-11-07 12:55:18 +00:00
{
int x, y;
png_bytep *row_pointers;
row_pointers= (png_bytep*)malloc(sizeof(png_bytep)*height);
for (y = 0; y < height; y++) {
row_pointers[y]= (png_bytep)malloc(sizeof(png_byte)*4*width);
for (x = 0; x < width; x++) {
char pixel = data[width*y+x];
row_pointers[y][x*4+0]= pixel;
row_pointers[y][x*4+1]= pixel;
row_pointers[y][x*4+2]= pixel;
row_pointers[y][x*4+3]= 255;
}
}
{
static int a = 0;
Camera tracking integration =========================== Commiting camera tracking integration gsoc project into trunk. This commit includes: - Bundled version of libmv library (with some changes against official repo, re-sync with libmv repo a bit later) - New datatype ID called MovieClip which is optimized to work with movie clips (both of movie files and image sequences) and doing camera/motion tracking operations. - New editor called Clip Editor which is currently used for motion/tracking stuff only, but which can be easily extended to work with masks too. This editor supports: * Loading movie files/image sequences * Build proxies with different size for loaded movie clip, also supports building undistorted proxies to increase speed of playback in undistorted mode. * Manual lens distortion mode calibration using grid and grease pencil * Supervised 2D tracking using two different algorithms KLT and SAD. * Basic algorithm for feature detection * Camera motion solving. scene orientation - New constraints to "link" scene objects with solved motions from clip: * Follow Track (make object follow 2D motion of track with given name or parent object to reconstructed 3D position of track) * Camera Solver to make camera moving in the same way as reconstructed camera This commit NOT includes changes from tomato branch: - New nodes (they'll be commited as separated patch) - Automatic image offset guessing for image input node and image editor (need to do more tests and gather more feedback) - Code cleanup in libmv-capi. It's not so critical cleanup, just increasing readability and understanadability of code. Better to make this chaneg when Keir will finish his current patch. More details about this project can be found on this page: http://wiki.blender.org/index.php/User:Nazg-gul/GSoC-2011 Further development of small features would be done in trunk, bigger/experimental features would first be implemented in tomato branch.
2011-11-07 12:55:18 +00:00
char buf[128];
snprintf(buf, sizeof(buf), "%s_%02d.png", prefix, ++a);
savePNGImage(row_pointers, width, height, 8, PNG_COLOR_TYPE_RGBA, buf);
}
for (y = 0; y < height; y++) {
free(row_pointers[y]);
}
free(row_pointers);
}
#endif
Planar tracking support for motion tracking =========================================== Major list of changes done in tomato branch: - Add a planar tracking implementation to libmv This adds a new planar tracking implementation to libmv. The tracker is based on Ceres[1], the new nonlinear minimizer that myself and Sameer released from Google as open source. Since the motion model is more involved, the interface is different than the RegionTracker interface used previously in Blender. The start of a C API in libmv-capi.{cpp,h} is also included. - Migrate from pat_{min,max} for markers to 4 corners representation Convert markers in the movie clip editor / 2D tracker from using pat_min and pat_max notation to using the a more general, 4-corner representation. There is still considerable porting work to do; in particular sliding from preview widget does not work correct for rotated markers. All other areas should be ported to new representation: * Added support of sliding individual corners. LMB slide + Ctrl would scale the whole pattern * S would scale the whole marker, S-S would scale pattern only * Added support of marker's rotation which is currently rotates only patterns around their centers or all markers around median, Rotation or other non-translation/scaling transformation of search area doesn't make sense. * Track Preview widget would display transformed pattern which libmv actually operates with. - "Efficient Second-order Minimization" for the planar tracker This implements the "Efficient Second-order Minimization" scheme, as supported by the existing translation tracker. This increases the amount of per-iteration work, but decreases the number of iterations required to converge and also increases the size of the basin of attraction for the optimization. - Remove the use of the legacy RegionTracker API from Blender, and replaces it with the new TrackRegion API. This also adds several features to the planar tracker in libmv: * Do a brute-force initialization of tracking similar to "Hybrid" mode in the stable release, but using all floats. This is slower but more accurate. It is still necessary to evaluate if the performance loss is worth it. In particular, this change is necessary to support high bit depth imagery. * Add support for masks over the search window. This is a step towards supporting user-defined tracker masks. The tracker masks will make it easy for users to make a mask for e.g. a ball. Not exposed into interface yet/ * Add Pearson product moment correlation coefficient checking (aka "Correlation" in the UI. This causes tracking failure if the tracked patch is not linearly related to the template. * Add support for warping a few points in addition to the supplied points. This is useful because the tracking code deliberately does not expose the underlying warp representation. Instead, warps are specified in an aparametric way via the correspondences. - Replace the old style tracker configuration panel with the new planar tracking panel. From a users perspective, this means: * The old "tracking algorithm" picker is gone. There is only 1 algorithm now. We may revisit this later, but I would much prefer to have only 1 algorithm. So far no optimization work has been done so the speed is not there yet. * There is now a dropdown to select the motion model. Choices: * Translation * Translation, rotation * Translation, scale * Translation, rotation, scale * Affine * Perspective * The old "Hybrid" mode is gone; instead there is a toggle to enable or disable translation-only tracker initialization. This is the equivalent of the hyrbid mode before, but rewritten to work with the new planar tracking modes. * The pyramid levels setting is gone. At a future date, the planar tracker will decide to use pyramids or not automatically. The pyramid setting was ultimately a mistake; with the brute force initialization it is unnecessary. - Add light-normalized tracking Added the ability to normalize patterns by their average value while tracking, to make them invariant to global illumination changes. Additional details could be found at wiki page [2] [1] http://code.google.com/p/ceres-solver [2] http://wiki.blender.org/index.php/Dev:Ref/Release_Notes/2.64/Motion_Tracker
2012-06-10 15:28:19 +00:00
/* ************ Planar tracker ************ */
/* TrackRegion (new planar tracker) */
int libmv_trackRegion(const struct libmv_trackRegionOptions *options,
const float *image1, int image1_width, int image1_height,
const float *image2, int image2_width, int image2_height,
Planar tracking support for motion tracking =========================================== Major list of changes done in tomato branch: - Add a planar tracking implementation to libmv This adds a new planar tracking implementation to libmv. The tracker is based on Ceres[1], the new nonlinear minimizer that myself and Sameer released from Google as open source. Since the motion model is more involved, the interface is different than the RegionTracker interface used previously in Blender. The start of a C API in libmv-capi.{cpp,h} is also included. - Migrate from pat_{min,max} for markers to 4 corners representation Convert markers in the movie clip editor / 2D tracker from using pat_min and pat_max notation to using the a more general, 4-corner representation. There is still considerable porting work to do; in particular sliding from preview widget does not work correct for rotated markers. All other areas should be ported to new representation: * Added support of sliding individual corners. LMB slide + Ctrl would scale the whole pattern * S would scale the whole marker, S-S would scale pattern only * Added support of marker's rotation which is currently rotates only patterns around their centers or all markers around median, Rotation or other non-translation/scaling transformation of search area doesn't make sense. * Track Preview widget would display transformed pattern which libmv actually operates with. - "Efficient Second-order Minimization" for the planar tracker This implements the "Efficient Second-order Minimization" scheme, as supported by the existing translation tracker. This increases the amount of per-iteration work, but decreases the number of iterations required to converge and also increases the size of the basin of attraction for the optimization. - Remove the use of the legacy RegionTracker API from Blender, and replaces it with the new TrackRegion API. This also adds several features to the planar tracker in libmv: * Do a brute-force initialization of tracking similar to "Hybrid" mode in the stable release, but using all floats. This is slower but more accurate. It is still necessary to evaluate if the performance loss is worth it. In particular, this change is necessary to support high bit depth imagery. * Add support for masks over the search window. This is a step towards supporting user-defined tracker masks. The tracker masks will make it easy for users to make a mask for e.g. a ball. Not exposed into interface yet/ * Add Pearson product moment correlation coefficient checking (aka "Correlation" in the UI. This causes tracking failure if the tracked patch is not linearly related to the template. * Add support for warping a few points in addition to the supplied points. This is useful because the tracking code deliberately does not expose the underlying warp representation. Instead, warps are specified in an aparametric way via the correspondences. - Replace the old style tracker configuration panel with the new planar tracking panel. From a users perspective, this means: * The old "tracking algorithm" picker is gone. There is only 1 algorithm now. We may revisit this later, but I would much prefer to have only 1 algorithm. So far no optimization work has been done so the speed is not there yet. * There is now a dropdown to select the motion model. Choices: * Translation * Translation, rotation * Translation, scale * Translation, rotation, scale * Affine * Perspective * The old "Hybrid" mode is gone; instead there is a toggle to enable or disable translation-only tracker initialization. This is the equivalent of the hyrbid mode before, but rewritten to work with the new planar tracking modes. * The pyramid levels setting is gone. At a future date, the planar tracker will decide to use pyramids or not automatically. The pyramid setting was ultimately a mistake; with the brute force initialization it is unnecessary. - Add light-normalized tracking Added the ability to normalize patterns by their average value while tracking, to make them invariant to global illumination changes. Additional details could be found at wiki page [2] [1] http://code.google.com/p/ceres-solver [2] http://wiki.blender.org/index.php/Dev:Ref/Release_Notes/2.64/Motion_Tracker
2012-06-10 15:28:19 +00:00
const double *x1, const double *y1,
struct libmv_trackRegionResult *result,
double *x2, double *y2)
{
double xx1[5], yy1[5];
double xx2[5], yy2[5];
bool tracking_result = false;
/* Convert to doubles for the libmv api. The four corners and the center. */
for (int i = 0; i < 5; ++i) {
xx1[i] = x1[i];
yy1[i] = y1[i];
xx2[i] = x2[i];
yy2[i] = y2[i];
}
libmv::TrackRegionOptions track_region_options;
libmv::FloatImage image1_mask;
Planar tracking support for motion tracking =========================================== Major list of changes done in tomato branch: - Add a planar tracking implementation to libmv This adds a new planar tracking implementation to libmv. The tracker is based on Ceres[1], the new nonlinear minimizer that myself and Sameer released from Google as open source. Since the motion model is more involved, the interface is different than the RegionTracker interface used previously in Blender. The start of a C API in libmv-capi.{cpp,h} is also included. - Migrate from pat_{min,max} for markers to 4 corners representation Convert markers in the movie clip editor / 2D tracker from using pat_min and pat_max notation to using the a more general, 4-corner representation. There is still considerable porting work to do; in particular sliding from preview widget does not work correct for rotated markers. All other areas should be ported to new representation: * Added support of sliding individual corners. LMB slide + Ctrl would scale the whole pattern * S would scale the whole marker, S-S would scale pattern only * Added support of marker's rotation which is currently rotates only patterns around their centers or all markers around median, Rotation or other non-translation/scaling transformation of search area doesn't make sense. * Track Preview widget would display transformed pattern which libmv actually operates with. - "Efficient Second-order Minimization" for the planar tracker This implements the "Efficient Second-order Minimization" scheme, as supported by the existing translation tracker. This increases the amount of per-iteration work, but decreases the number of iterations required to converge and also increases the size of the basin of attraction for the optimization. - Remove the use of the legacy RegionTracker API from Blender, and replaces it with the new TrackRegion API. This also adds several features to the planar tracker in libmv: * Do a brute-force initialization of tracking similar to "Hybrid" mode in the stable release, but using all floats. This is slower but more accurate. It is still necessary to evaluate if the performance loss is worth it. In particular, this change is necessary to support high bit depth imagery. * Add support for masks over the search window. This is a step towards supporting user-defined tracker masks. The tracker masks will make it easy for users to make a mask for e.g. a ball. Not exposed into interface yet/ * Add Pearson product moment correlation coefficient checking (aka "Correlation" in the UI. This causes tracking failure if the tracked patch is not linearly related to the template. * Add support for warping a few points in addition to the supplied points. This is useful because the tracking code deliberately does not expose the underlying warp representation. Instead, warps are specified in an aparametric way via the correspondences. - Replace the old style tracker configuration panel with the new planar tracking panel. From a users perspective, this means: * The old "tracking algorithm" picker is gone. There is only 1 algorithm now. We may revisit this later, but I would much prefer to have only 1 algorithm. So far no optimization work has been done so the speed is not there yet. * There is now a dropdown to select the motion model. Choices: * Translation * Translation, rotation * Translation, scale * Translation, rotation, scale * Affine * Perspective * The old "Hybrid" mode is gone; instead there is a toggle to enable or disable translation-only tracker initialization. This is the equivalent of the hyrbid mode before, but rewritten to work with the new planar tracking modes. * The pyramid levels setting is gone. At a future date, the planar tracker will decide to use pyramids or not automatically. The pyramid setting was ultimately a mistake; with the brute force initialization it is unnecessary. - Add light-normalized tracking Added the ability to normalize patterns by their average value while tracking, to make them invariant to global illumination changes. Additional details could be found at wiki page [2] [1] http://code.google.com/p/ceres-solver [2] http://wiki.blender.org/index.php/Dev:Ref/Release_Notes/2.64/Motion_Tracker
2012-06-10 15:28:19 +00:00
switch (options->motion_model) {
#define LIBMV_CONVERT(the_model) \
case libmv::TrackRegionOptions::the_model: \
track_region_options.mode = libmv::TrackRegionOptions::the_model; \
break;
LIBMV_CONVERT(TRANSLATION)
LIBMV_CONVERT(TRANSLATION_ROTATION)
LIBMV_CONVERT(TRANSLATION_SCALE)
LIBMV_CONVERT(TRANSLATION_ROTATION_SCALE)
LIBMV_CONVERT(AFFINE)
LIBMV_CONVERT(HOMOGRAPHY)
#undef LIBMV_CONVERT
}
track_region_options.minimum_correlation = options->minimum_correlation;
track_region_options.max_iterations = options->num_iterations;
track_region_options.sigma = options->sigma;
track_region_options.num_extra_points = 1;
track_region_options.image1_mask = NULL;
track_region_options.use_brute_initialization = options->use_brute;
track_region_options.use_normalized_intensities = options->use_normalization;
if (options->image1_mask) {
floatBufToImage(options->image1_mask, image1_width, image1_height, 1, &image1_mask);
track_region_options.image1_mask = &image1_mask;
}
Planar tracking support for motion tracking =========================================== Major list of changes done in tomato branch: - Add a planar tracking implementation to libmv This adds a new planar tracking implementation to libmv. The tracker is based on Ceres[1], the new nonlinear minimizer that myself and Sameer released from Google as open source. Since the motion model is more involved, the interface is different than the RegionTracker interface used previously in Blender. The start of a C API in libmv-capi.{cpp,h} is also included. - Migrate from pat_{min,max} for markers to 4 corners representation Convert markers in the movie clip editor / 2D tracker from using pat_min and pat_max notation to using the a more general, 4-corner representation. There is still considerable porting work to do; in particular sliding from preview widget does not work correct for rotated markers. All other areas should be ported to new representation: * Added support of sliding individual corners. LMB slide + Ctrl would scale the whole pattern * S would scale the whole marker, S-S would scale pattern only * Added support of marker's rotation which is currently rotates only patterns around their centers or all markers around median, Rotation or other non-translation/scaling transformation of search area doesn't make sense. * Track Preview widget would display transformed pattern which libmv actually operates with. - "Efficient Second-order Minimization" for the planar tracker This implements the "Efficient Second-order Minimization" scheme, as supported by the existing translation tracker. This increases the amount of per-iteration work, but decreases the number of iterations required to converge and also increases the size of the basin of attraction for the optimization. - Remove the use of the legacy RegionTracker API from Blender, and replaces it with the new TrackRegion API. This also adds several features to the planar tracker in libmv: * Do a brute-force initialization of tracking similar to "Hybrid" mode in the stable release, but using all floats. This is slower but more accurate. It is still necessary to evaluate if the performance loss is worth it. In particular, this change is necessary to support high bit depth imagery. * Add support for masks over the search window. This is a step towards supporting user-defined tracker masks. The tracker masks will make it easy for users to make a mask for e.g. a ball. Not exposed into interface yet/ * Add Pearson product moment correlation coefficient checking (aka "Correlation" in the UI. This causes tracking failure if the tracked patch is not linearly related to the template. * Add support for warping a few points in addition to the supplied points. This is useful because the tracking code deliberately does not expose the underlying warp representation. Instead, warps are specified in an aparametric way via the correspondences. - Replace the old style tracker configuration panel with the new planar tracking panel. From a users perspective, this means: * The old "tracking algorithm" picker is gone. There is only 1 algorithm now. We may revisit this later, but I would much prefer to have only 1 algorithm. So far no optimization work has been done so the speed is not there yet. * There is now a dropdown to select the motion model. Choices: * Translation * Translation, rotation * Translation, scale * Translation, rotation, scale * Affine * Perspective * The old "Hybrid" mode is gone; instead there is a toggle to enable or disable translation-only tracker initialization. This is the equivalent of the hyrbid mode before, but rewritten to work with the new planar tracking modes. * The pyramid levels setting is gone. At a future date, the planar tracker will decide to use pyramids or not automatically. The pyramid setting was ultimately a mistake; with the brute force initialization it is unnecessary. - Add light-normalized tracking Added the ability to normalize patterns by their average value while tracking, to make them invariant to global illumination changes. Additional details could be found at wiki page [2] [1] http://code.google.com/p/ceres-solver [2] http://wiki.blender.org/index.php/Dev:Ref/Release_Notes/2.64/Motion_Tracker
2012-06-10 15:28:19 +00:00
/* Convert from raw float buffers to libmv's FloatImage. */
libmv::FloatImage old_patch, new_patch;
floatBufToImage(image1, image1_width, image1_height, 1, &old_patch);
floatBufToImage(image2, image2_width, image2_height, 1, &new_patch);
Planar tracking support for motion tracking =========================================== Major list of changes done in tomato branch: - Add a planar tracking implementation to libmv This adds a new planar tracking implementation to libmv. The tracker is based on Ceres[1], the new nonlinear minimizer that myself and Sameer released from Google as open source. Since the motion model is more involved, the interface is different than the RegionTracker interface used previously in Blender. The start of a C API in libmv-capi.{cpp,h} is also included. - Migrate from pat_{min,max} for markers to 4 corners representation Convert markers in the movie clip editor / 2D tracker from using pat_min and pat_max notation to using the a more general, 4-corner representation. There is still considerable porting work to do; in particular sliding from preview widget does not work correct for rotated markers. All other areas should be ported to new representation: * Added support of sliding individual corners. LMB slide + Ctrl would scale the whole pattern * S would scale the whole marker, S-S would scale pattern only * Added support of marker's rotation which is currently rotates only patterns around their centers or all markers around median, Rotation or other non-translation/scaling transformation of search area doesn't make sense. * Track Preview widget would display transformed pattern which libmv actually operates with. - "Efficient Second-order Minimization" for the planar tracker This implements the "Efficient Second-order Minimization" scheme, as supported by the existing translation tracker. This increases the amount of per-iteration work, but decreases the number of iterations required to converge and also increases the size of the basin of attraction for the optimization. - Remove the use of the legacy RegionTracker API from Blender, and replaces it with the new TrackRegion API. This also adds several features to the planar tracker in libmv: * Do a brute-force initialization of tracking similar to "Hybrid" mode in the stable release, but using all floats. This is slower but more accurate. It is still necessary to evaluate if the performance loss is worth it. In particular, this change is necessary to support high bit depth imagery. * Add support for masks over the search window. This is a step towards supporting user-defined tracker masks. The tracker masks will make it easy for users to make a mask for e.g. a ball. Not exposed into interface yet/ * Add Pearson product moment correlation coefficient checking (aka "Correlation" in the UI. This causes tracking failure if the tracked patch is not linearly related to the template. * Add support for warping a few points in addition to the supplied points. This is useful because the tracking code deliberately does not expose the underlying warp representation. Instead, warps are specified in an aparametric way via the correspondences. - Replace the old style tracker configuration panel with the new planar tracking panel. From a users perspective, this means: * The old "tracking algorithm" picker is gone. There is only 1 algorithm now. We may revisit this later, but I would much prefer to have only 1 algorithm. So far no optimization work has been done so the speed is not there yet. * There is now a dropdown to select the motion model. Choices: * Translation * Translation, rotation * Translation, scale * Translation, rotation, scale * Affine * Perspective * The old "Hybrid" mode is gone; instead there is a toggle to enable or disable translation-only tracker initialization. This is the equivalent of the hyrbid mode before, but rewritten to work with the new planar tracking modes. * The pyramid levels setting is gone. At a future date, the planar tracker will decide to use pyramids or not automatically. The pyramid setting was ultimately a mistake; with the brute force initialization it is unnecessary. - Add light-normalized tracking Added the ability to normalize patterns by their average value while tracking, to make them invariant to global illumination changes. Additional details could be found at wiki page [2] [1] http://code.google.com/p/ceres-solver [2] http://wiki.blender.org/index.php/Dev:Ref/Release_Notes/2.64/Motion_Tracker
2012-06-10 15:28:19 +00:00
libmv::TrackRegionResult track_region_result;
libmv::TrackRegion(old_patch, new_patch, xx1, yy1, track_region_options, xx2, yy2, &track_region_result);
/* Convert to floats for the blender api. */
for (int i = 0; i < 5; ++i) {
x2[i] = xx2[i];
y2[i] = yy2[i];
}
/* TODO(keir): Update the termination string with failure details. */
if (track_region_result.termination == libmv::TrackRegionResult::PARAMETER_TOLERANCE ||
track_region_result.termination == libmv::TrackRegionResult::FUNCTION_TOLERANCE ||
track_region_result.termination == libmv::TrackRegionResult::GRADIENT_TOLERANCE ||
track_region_result.termination == libmv::TrackRegionResult::NO_CONVERGENCE)
{
tracking_result = true;
}
#if defined(DUMP_FAILURE) || defined(DUMP_ALWAYS)
#if defined(DUMP_ALWAYS)
{
#else
if (!tracking_result) {
#endif
saveImage("old_patch", old_patch, x1[4], y1[4]);
saveImage("new_patch", new_patch, x2[4], y2[4]);
if (options->image1_mask)
saveImage("mask", image1_mask, x2[4], y2[4]);
Planar tracking support for motion tracking =========================================== Major list of changes done in tomato branch: - Add a planar tracking implementation to libmv This adds a new planar tracking implementation to libmv. The tracker is based on Ceres[1], the new nonlinear minimizer that myself and Sameer released from Google as open source. Since the motion model is more involved, the interface is different than the RegionTracker interface used previously in Blender. The start of a C API in libmv-capi.{cpp,h} is also included. - Migrate from pat_{min,max} for markers to 4 corners representation Convert markers in the movie clip editor / 2D tracker from using pat_min and pat_max notation to using the a more general, 4-corner representation. There is still considerable porting work to do; in particular sliding from preview widget does not work correct for rotated markers. All other areas should be ported to new representation: * Added support of sliding individual corners. LMB slide + Ctrl would scale the whole pattern * S would scale the whole marker, S-S would scale pattern only * Added support of marker's rotation which is currently rotates only patterns around their centers or all markers around median, Rotation or other non-translation/scaling transformation of search area doesn't make sense. * Track Preview widget would display transformed pattern which libmv actually operates with. - "Efficient Second-order Minimization" for the planar tracker This implements the "Efficient Second-order Minimization" scheme, as supported by the existing translation tracker. This increases the amount of per-iteration work, but decreases the number of iterations required to converge and also increases the size of the basin of attraction for the optimization. - Remove the use of the legacy RegionTracker API from Blender, and replaces it with the new TrackRegion API. This also adds several features to the planar tracker in libmv: * Do a brute-force initialization of tracking similar to "Hybrid" mode in the stable release, but using all floats. This is slower but more accurate. It is still necessary to evaluate if the performance loss is worth it. In particular, this change is necessary to support high bit depth imagery. * Add support for masks over the search window. This is a step towards supporting user-defined tracker masks. The tracker masks will make it easy for users to make a mask for e.g. a ball. Not exposed into interface yet/ * Add Pearson product moment correlation coefficient checking (aka "Correlation" in the UI. This causes tracking failure if the tracked patch is not linearly related to the template. * Add support for warping a few points in addition to the supplied points. This is useful because the tracking code deliberately does not expose the underlying warp representation. Instead, warps are specified in an aparametric way via the correspondences. - Replace the old style tracker configuration panel with the new planar tracking panel. From a users perspective, this means: * The old "tracking algorithm" picker is gone. There is only 1 algorithm now. We may revisit this later, but I would much prefer to have only 1 algorithm. So far no optimization work has been done so the speed is not there yet. * There is now a dropdown to select the motion model. Choices: * Translation * Translation, rotation * Translation, scale * Translation, rotation, scale * Affine * Perspective * The old "Hybrid" mode is gone; instead there is a toggle to enable or disable translation-only tracker initialization. This is the equivalent of the hyrbid mode before, but rewritten to work with the new planar tracking modes. * The pyramid levels setting is gone. At a future date, the planar tracker will decide to use pyramids or not automatically. The pyramid setting was ultimately a mistake; with the brute force initialization it is unnecessary. - Add light-normalized tracking Added the ability to normalize patterns by their average value while tracking, to make them invariant to global illumination changes. Additional details could be found at wiki page [2] [1] http://code.google.com/p/ceres-solver [2] http://wiki.blender.org/index.php/Dev:Ref/Release_Notes/2.64/Motion_Tracker
2012-06-10 15:28:19 +00:00
}
#endif
return tracking_result;
}
void libmv_samplePlanarPatch(const float *image, int width, int height,
int channels, const double *xs, const double *ys,
int num_samples_x, int num_samples_y,
const float *mask, float *patch,
Planar tracking support for motion tracking =========================================== Major list of changes done in tomato branch: - Add a planar tracking implementation to libmv This adds a new planar tracking implementation to libmv. The tracker is based on Ceres[1], the new nonlinear minimizer that myself and Sameer released from Google as open source. Since the motion model is more involved, the interface is different than the RegionTracker interface used previously in Blender. The start of a C API in libmv-capi.{cpp,h} is also included. - Migrate from pat_{min,max} for markers to 4 corners representation Convert markers in the movie clip editor / 2D tracker from using pat_min and pat_max notation to using the a more general, 4-corner representation. There is still considerable porting work to do; in particular sliding from preview widget does not work correct for rotated markers. All other areas should be ported to new representation: * Added support of sliding individual corners. LMB slide + Ctrl would scale the whole pattern * S would scale the whole marker, S-S would scale pattern only * Added support of marker's rotation which is currently rotates only patterns around their centers or all markers around median, Rotation or other non-translation/scaling transformation of search area doesn't make sense. * Track Preview widget would display transformed pattern which libmv actually operates with. - "Efficient Second-order Minimization" for the planar tracker This implements the "Efficient Second-order Minimization" scheme, as supported by the existing translation tracker. This increases the amount of per-iteration work, but decreases the number of iterations required to converge and also increases the size of the basin of attraction for the optimization. - Remove the use of the legacy RegionTracker API from Blender, and replaces it with the new TrackRegion API. This also adds several features to the planar tracker in libmv: * Do a brute-force initialization of tracking similar to "Hybrid" mode in the stable release, but using all floats. This is slower but more accurate. It is still necessary to evaluate if the performance loss is worth it. In particular, this change is necessary to support high bit depth imagery. * Add support for masks over the search window. This is a step towards supporting user-defined tracker masks. The tracker masks will make it easy for users to make a mask for e.g. a ball. Not exposed into interface yet/ * Add Pearson product moment correlation coefficient checking (aka "Correlation" in the UI. This causes tracking failure if the tracked patch is not linearly related to the template. * Add support for warping a few points in addition to the supplied points. This is useful because the tracking code deliberately does not expose the underlying warp representation. Instead, warps are specified in an aparametric way via the correspondences. - Replace the old style tracker configuration panel with the new planar tracking panel. From a users perspective, this means: * The old "tracking algorithm" picker is gone. There is only 1 algorithm now. We may revisit this later, but I would much prefer to have only 1 algorithm. So far no optimization work has been done so the speed is not there yet. * There is now a dropdown to select the motion model. Choices: * Translation * Translation, rotation * Translation, scale * Translation, rotation, scale * Affine * Perspective * The old "Hybrid" mode is gone; instead there is a toggle to enable or disable translation-only tracker initialization. This is the equivalent of the hyrbid mode before, but rewritten to work with the new planar tracking modes. * The pyramid levels setting is gone. At a future date, the planar tracker will decide to use pyramids or not automatically. The pyramid setting was ultimately a mistake; with the brute force initialization it is unnecessary. - Add light-normalized tracking Added the ability to normalize patterns by their average value while tracking, to make them invariant to global illumination changes. Additional details could be found at wiki page [2] [1] http://code.google.com/p/ceres-solver [2] http://wiki.blender.org/index.php/Dev:Ref/Release_Notes/2.64/Motion_Tracker
2012-06-10 15:28:19 +00:00
double *warped_position_x, double *warped_position_y)
{
libmv::FloatImage libmv_image, libmv_patch, libmv_mask;
libmv::FloatImage *libmv_mask_for_sample = NULL;
Planar tracking support for motion tracking =========================================== Major list of changes done in tomato branch: - Add a planar tracking implementation to libmv This adds a new planar tracking implementation to libmv. The tracker is based on Ceres[1], the new nonlinear minimizer that myself and Sameer released from Google as open source. Since the motion model is more involved, the interface is different than the RegionTracker interface used previously in Blender. The start of a C API in libmv-capi.{cpp,h} is also included. - Migrate from pat_{min,max} for markers to 4 corners representation Convert markers in the movie clip editor / 2D tracker from using pat_min and pat_max notation to using the a more general, 4-corner representation. There is still considerable porting work to do; in particular sliding from preview widget does not work correct for rotated markers. All other areas should be ported to new representation: * Added support of sliding individual corners. LMB slide + Ctrl would scale the whole pattern * S would scale the whole marker, S-S would scale pattern only * Added support of marker's rotation which is currently rotates only patterns around their centers or all markers around median, Rotation or other non-translation/scaling transformation of search area doesn't make sense. * Track Preview widget would display transformed pattern which libmv actually operates with. - "Efficient Second-order Minimization" for the planar tracker This implements the "Efficient Second-order Minimization" scheme, as supported by the existing translation tracker. This increases the amount of per-iteration work, but decreases the number of iterations required to converge and also increases the size of the basin of attraction for the optimization. - Remove the use of the legacy RegionTracker API from Blender, and replaces it with the new TrackRegion API. This also adds several features to the planar tracker in libmv: * Do a brute-force initialization of tracking similar to "Hybrid" mode in the stable release, but using all floats. This is slower but more accurate. It is still necessary to evaluate if the performance loss is worth it. In particular, this change is necessary to support high bit depth imagery. * Add support for masks over the search window. This is a step towards supporting user-defined tracker masks. The tracker masks will make it easy for users to make a mask for e.g. a ball. Not exposed into interface yet/ * Add Pearson product moment correlation coefficient checking (aka "Correlation" in the UI. This causes tracking failure if the tracked patch is not linearly related to the template. * Add support for warping a few points in addition to the supplied points. This is useful because the tracking code deliberately does not expose the underlying warp representation. Instead, warps are specified in an aparametric way via the correspondences. - Replace the old style tracker configuration panel with the new planar tracking panel. From a users perspective, this means: * The old "tracking algorithm" picker is gone. There is only 1 algorithm now. We may revisit this later, but I would much prefer to have only 1 algorithm. So far no optimization work has been done so the speed is not there yet. * There is now a dropdown to select the motion model. Choices: * Translation * Translation, rotation * Translation, scale * Translation, rotation, scale * Affine * Perspective * The old "Hybrid" mode is gone; instead there is a toggle to enable or disable translation-only tracker initialization. This is the equivalent of the hyrbid mode before, but rewritten to work with the new planar tracking modes. * The pyramid levels setting is gone. At a future date, the planar tracker will decide to use pyramids or not automatically. The pyramid setting was ultimately a mistake; with the brute force initialization it is unnecessary. - Add light-normalized tracking Added the ability to normalize patterns by their average value while tracking, to make them invariant to global illumination changes. Additional details could be found at wiki page [2] [1] http://code.google.com/p/ceres-solver [2] http://wiki.blender.org/index.php/Dev:Ref/Release_Notes/2.64/Motion_Tracker
2012-06-10 15:28:19 +00:00
floatBufToImage(image, width, height, channels, &libmv_image);
if (mask) {
floatBufToImage(mask, width, height, 1, &libmv_mask);
libmv_mask_for_sample = &libmv_mask;
}
Planar tracking support for motion tracking =========================================== Major list of changes done in tomato branch: - Add a planar tracking implementation to libmv This adds a new planar tracking implementation to libmv. The tracker is based on Ceres[1], the new nonlinear minimizer that myself and Sameer released from Google as open source. Since the motion model is more involved, the interface is different than the RegionTracker interface used previously in Blender. The start of a C API in libmv-capi.{cpp,h} is also included. - Migrate from pat_{min,max} for markers to 4 corners representation Convert markers in the movie clip editor / 2D tracker from using pat_min and pat_max notation to using the a more general, 4-corner representation. There is still considerable porting work to do; in particular sliding from preview widget does not work correct for rotated markers. All other areas should be ported to new representation: * Added support of sliding individual corners. LMB slide + Ctrl would scale the whole pattern * S would scale the whole marker, S-S would scale pattern only * Added support of marker's rotation which is currently rotates only patterns around their centers or all markers around median, Rotation or other non-translation/scaling transformation of search area doesn't make sense. * Track Preview widget would display transformed pattern which libmv actually operates with. - "Efficient Second-order Minimization" for the planar tracker This implements the "Efficient Second-order Minimization" scheme, as supported by the existing translation tracker. This increases the amount of per-iteration work, but decreases the number of iterations required to converge and also increases the size of the basin of attraction for the optimization. - Remove the use of the legacy RegionTracker API from Blender, and replaces it with the new TrackRegion API. This also adds several features to the planar tracker in libmv: * Do a brute-force initialization of tracking similar to "Hybrid" mode in the stable release, but using all floats. This is slower but more accurate. It is still necessary to evaluate if the performance loss is worth it. In particular, this change is necessary to support high bit depth imagery. * Add support for masks over the search window. This is a step towards supporting user-defined tracker masks. The tracker masks will make it easy for users to make a mask for e.g. a ball. Not exposed into interface yet/ * Add Pearson product moment correlation coefficient checking (aka "Correlation" in the UI. This causes tracking failure if the tracked patch is not linearly related to the template. * Add support for warping a few points in addition to the supplied points. This is useful because the tracking code deliberately does not expose the underlying warp representation. Instead, warps are specified in an aparametric way via the correspondences. - Replace the old style tracker configuration panel with the new planar tracking panel. From a users perspective, this means: * The old "tracking algorithm" picker is gone. There is only 1 algorithm now. We may revisit this later, but I would much prefer to have only 1 algorithm. So far no optimization work has been done so the speed is not there yet. * There is now a dropdown to select the motion model. Choices: * Translation * Translation, rotation * Translation, scale * Translation, rotation, scale * Affine * Perspective * The old "Hybrid" mode is gone; instead there is a toggle to enable or disable translation-only tracker initialization. This is the equivalent of the hyrbid mode before, but rewritten to work with the new planar tracking modes. * The pyramid levels setting is gone. At a future date, the planar tracker will decide to use pyramids or not automatically. The pyramid setting was ultimately a mistake; with the brute force initialization it is unnecessary. - Add light-normalized tracking Added the ability to normalize patterns by their average value while tracking, to make them invariant to global illumination changes. Additional details could be found at wiki page [2] [1] http://code.google.com/p/ceres-solver [2] http://wiki.blender.org/index.php/Dev:Ref/Release_Notes/2.64/Motion_Tracker
2012-06-10 15:28:19 +00:00
libmv::SamplePlanarPatch(libmv_image, xs, ys, num_samples_x, num_samples_y,
libmv_mask_for_sample, &libmv_patch,
warped_position_x, warped_position_y);
Planar tracking support for motion tracking =========================================== Major list of changes done in tomato branch: - Add a planar tracking implementation to libmv This adds a new planar tracking implementation to libmv. The tracker is based on Ceres[1], the new nonlinear minimizer that myself and Sameer released from Google as open source. Since the motion model is more involved, the interface is different than the RegionTracker interface used previously in Blender. The start of a C API in libmv-capi.{cpp,h} is also included. - Migrate from pat_{min,max} for markers to 4 corners representation Convert markers in the movie clip editor / 2D tracker from using pat_min and pat_max notation to using the a more general, 4-corner representation. There is still considerable porting work to do; in particular sliding from preview widget does not work correct for rotated markers. All other areas should be ported to new representation: * Added support of sliding individual corners. LMB slide + Ctrl would scale the whole pattern * S would scale the whole marker, S-S would scale pattern only * Added support of marker's rotation which is currently rotates only patterns around their centers or all markers around median, Rotation or other non-translation/scaling transformation of search area doesn't make sense. * Track Preview widget would display transformed pattern which libmv actually operates with. - "Efficient Second-order Minimization" for the planar tracker This implements the "Efficient Second-order Minimization" scheme, as supported by the existing translation tracker. This increases the amount of per-iteration work, but decreases the number of iterations required to converge and also increases the size of the basin of attraction for the optimization. - Remove the use of the legacy RegionTracker API from Blender, and replaces it with the new TrackRegion API. This also adds several features to the planar tracker in libmv: * Do a brute-force initialization of tracking similar to "Hybrid" mode in the stable release, but using all floats. This is slower but more accurate. It is still necessary to evaluate if the performance loss is worth it. In particular, this change is necessary to support high bit depth imagery. * Add support for masks over the search window. This is a step towards supporting user-defined tracker masks. The tracker masks will make it easy for users to make a mask for e.g. a ball. Not exposed into interface yet/ * Add Pearson product moment correlation coefficient checking (aka "Correlation" in the UI. This causes tracking failure if the tracked patch is not linearly related to the template. * Add support for warping a few points in addition to the supplied points. This is useful because the tracking code deliberately does not expose the underlying warp representation. Instead, warps are specified in an aparametric way via the correspondences. - Replace the old style tracker configuration panel with the new planar tracking panel. From a users perspective, this means: * The old "tracking algorithm" picker is gone. There is only 1 algorithm now. We may revisit this later, but I would much prefer to have only 1 algorithm. So far no optimization work has been done so the speed is not there yet. * There is now a dropdown to select the motion model. Choices: * Translation * Translation, rotation * Translation, scale * Translation, rotation, scale * Affine * Perspective * The old "Hybrid" mode is gone; instead there is a toggle to enable or disable translation-only tracker initialization. This is the equivalent of the hyrbid mode before, but rewritten to work with the new planar tracking modes. * The pyramid levels setting is gone. At a future date, the planar tracker will decide to use pyramids or not automatically. The pyramid setting was ultimately a mistake; with the brute force initialization it is unnecessary. - Add light-normalized tracking Added the ability to normalize patterns by their average value while tracking, to make them invariant to global illumination changes. Additional details could be found at wiki page [2] [1] http://code.google.com/p/ceres-solver [2] http://wiki.blender.org/index.php/Dev:Ref/Release_Notes/2.64/Motion_Tracker
2012-06-10 15:28:19 +00:00
imageToFloatBuf(&libmv_patch, channels, patch);
}
Camera tracking integration =========================== Commiting camera tracking integration gsoc project into trunk. This commit includes: - Bundled version of libmv library (with some changes against official repo, re-sync with libmv repo a bit later) - New datatype ID called MovieClip which is optimized to work with movie clips (both of movie files and image sequences) and doing camera/motion tracking operations. - New editor called Clip Editor which is currently used for motion/tracking stuff only, but which can be easily extended to work with masks too. This editor supports: * Loading movie files/image sequences * Build proxies with different size for loaded movie clip, also supports building undistorted proxies to increase speed of playback in undistorted mode. * Manual lens distortion mode calibration using grid and grease pencil * Supervised 2D tracking using two different algorithms KLT and SAD. * Basic algorithm for feature detection * Camera motion solving. scene orientation - New constraints to "link" scene objects with solved motions from clip: * Follow Track (make object follow 2D motion of track with given name or parent object to reconstructed 3D position of track) * Camera Solver to make camera moving in the same way as reconstructed camera This commit NOT includes changes from tomato branch: - New nodes (they'll be commited as separated patch) - Automatic image offset guessing for image input node and image editor (need to do more tests and gather more feedback) - Code cleanup in libmv-capi. It's not so critical cleanup, just increasing readability and understanadability of code. Better to make this chaneg when Keir will finish his current patch. More details about this project can be found on this page: http://wiki.blender.org/index.php/User:Nazg-gul/GSoC-2011 Further development of small features would be done in trunk, bigger/experimental features would first be implemented in tomato branch.
2011-11-07 12:55:18 +00:00
/* ************ Tracks ************ */
libmv_Tracks *libmv_tracksNew(void)
{
libmv::Tracks *libmv_tracks = new libmv::Tracks();
return (libmv_Tracks *)libmv_tracks;
}
void libmv_tracksInsert(struct libmv_Tracks *libmv_tracks, int image, int track, double x, double y)
{
((libmv::Tracks*)libmv_tracks)->Insert(image, track, x, y);
}
void libmv_tracksDestroy(libmv_Tracks *libmv_tracks)
{
delete (libmv::Tracks*)libmv_tracks;
}
/* ************ Reconstruction solver ************ */
class ReconstructUpdateCallback : public libmv::ProgressUpdateCallback {
public:
ReconstructUpdateCallback(reconstruct_progress_update_cb progress_update_callback,
void *callback_customdata)
{
progress_update_callback_ = progress_update_callback;
callback_customdata_ = callback_customdata;
}
void invoke(double progress, const char *message)
{
if(progress_update_callback_) {
progress_update_callback_(callback_customdata_, progress, message);
}
}
protected:
reconstruct_progress_update_cb progress_update_callback_;
void *callback_customdata_;
};
Assorted camera tracker improvements - Add support for refining the camera's intrinsic parameters during a solve. Currently, refining supports only the following combinations of intrinsic parameters: f f, cx, cy f, cx, cy, k1, k2 f, k1 f, k1, k2 This is not the same as autocalibration, since the user must still make a reasonable initial guess about the focal length and other parameters, whereas true autocalibration would eliminate the need for the user specify intrinsic parameters at all. However, the solver works well with only rough guesses for the focal length, so perhaps full autocalibation is not that important. Adding support for the last two combinations, (f, k1) and (f, k1, k2) required changes to the library libmv depends on for bundle adjustment, SSBA. These changes should get ported upstream not just to libmv but to SSBA as well. - Improved the region of convergence for bundle adjustment by increasing the number of Levenberg-Marquardt iterations from 50 to 500. This way, the solver is able to crawl out of the bad local minima it gets stuck in when changing from, for example, bundling k1 and k2 to just k1 and resetting k2 to 0. - Add several new region tracker implementations. A region tracker is a libmv concept, which refers to tracking a template image pattern through frames. The impact to end users is that tracking should "just work better". I am reserving a more detailed writeup, and maybe a paper, for later. - Other libmv tweaks, such as detecting that a tracker is headed outside of the image bounds. This includes several changes made directly to the libmv extern code rather expecting to get those changes through normal libmv channels, because I, the libmv BDFL, decided it was faster to work on libmv directly in Blender, then later reverse-port the libmv changes from Blender back into libmv trunk. The interesting part is that I added a full Levenberg-Marquardt loop to the region tracking code, which should lead to a more stable solutions. I also added a hacky implementation of "Efficient Second-Order Minimization" for tracking, which works nicely. A more detailed quantitative evaluation will follow. Original patch by Keir, cleaned a bit by myself.
2011-11-14 06:41:23 +00:00
int libmv_refineParametersAreValid(int parameters) {
return (parameters == (LIBMV_REFINE_FOCAL_LENGTH)) ||
(parameters == (LIBMV_REFINE_FOCAL_LENGTH |
LIBMV_REFINE_PRINCIPAL_POINT)) ||
(parameters == (LIBMV_REFINE_FOCAL_LENGTH |
LIBMV_REFINE_PRINCIPAL_POINT |
LIBMV_REFINE_RADIAL_DISTORTION_K1 |
LIBMV_REFINE_RADIAL_DISTORTION_K2)) ||
(parameters == (LIBMV_REFINE_FOCAL_LENGTH |
LIBMV_REFINE_RADIAL_DISTORTION_K1 |
LIBMV_REFINE_RADIAL_DISTORTION_K2)) ||
(parameters == (LIBMV_REFINE_FOCAL_LENGTH |
LIBMV_REFINE_RADIAL_DISTORTION_K1)) ||
(parameters == (LIBMV_REFINE_RADIAL_DISTORTION_K1 |
LIBMV_REFINE_RADIAL_DISTORTION_K2));
Assorted camera tracker improvements - Add support for refining the camera's intrinsic parameters during a solve. Currently, refining supports only the following combinations of intrinsic parameters: f f, cx, cy f, cx, cy, k1, k2 f, k1 f, k1, k2 This is not the same as autocalibration, since the user must still make a reasonable initial guess about the focal length and other parameters, whereas true autocalibration would eliminate the need for the user specify intrinsic parameters at all. However, the solver works well with only rough guesses for the focal length, so perhaps full autocalibation is not that important. Adding support for the last two combinations, (f, k1) and (f, k1, k2) required changes to the library libmv depends on for bundle adjustment, SSBA. These changes should get ported upstream not just to libmv but to SSBA as well. - Improved the region of convergence for bundle adjustment by increasing the number of Levenberg-Marquardt iterations from 50 to 500. This way, the solver is able to crawl out of the bad local minima it gets stuck in when changing from, for example, bundling k1 and k2 to just k1 and resetting k2 to 0. - Add several new region tracker implementations. A region tracker is a libmv concept, which refers to tracking a template image pattern through frames. The impact to end users is that tracking should "just work better". I am reserving a more detailed writeup, and maybe a paper, for later. - Other libmv tweaks, such as detecting that a tracker is headed outside of the image bounds. This includes several changes made directly to the libmv extern code rather expecting to get those changes through normal libmv channels, because I, the libmv BDFL, decided it was faster to work on libmv directly in Blender, then later reverse-port the libmv changes from Blender back into libmv trunk. The interesting part is that I added a full Levenberg-Marquardt loop to the region tracking code, which should lead to a more stable solutions. I also added a hacky implementation of "Efficient Second-Order Minimization" for tracking, which works nicely. A more detailed quantitative evaluation will follow. Original patch by Keir, cleaned a bit by myself.
2011-11-14 06:41:23 +00:00
}
static void libmv_solveRefineIntrinsics(libmv::Tracks *tracks, libmv::CameraIntrinsics *intrinsics,
libmv::EuclideanReconstruction *reconstruction, int refine_intrinsics,
reconstruct_progress_update_cb progress_update_callback, void *callback_customdata,
int bundle_constraints = libmv::BUNDLE_NO_CONSTRAINTS)
{
/* only a few combinations are supported but trust the caller */
int libmv_refine_flags = 0;
if (refine_intrinsics & LIBMV_REFINE_FOCAL_LENGTH) {
libmv_refine_flags |= libmv::BUNDLE_FOCAL_LENGTH;
}
if (refine_intrinsics & LIBMV_REFINE_PRINCIPAL_POINT) {
libmv_refine_flags |= libmv::BUNDLE_PRINCIPAL_POINT;
}
if (refine_intrinsics & LIBMV_REFINE_RADIAL_DISTORTION_K1) {
libmv_refine_flags |= libmv::BUNDLE_RADIAL_K1;
}
if (refine_intrinsics & LIBMV_REFINE_RADIAL_DISTORTION_K2) {
libmv_refine_flags |= libmv::BUNDLE_RADIAL_K2;
}
progress_update_callback(callback_customdata, 1.0, "Refining solution");
libmv::EuclideanBundleCommonIntrinsics(*(libmv::Tracks *)tracks, libmv_refine_flags,
reconstruction, intrinsics, bundle_constraints);
}
Assorted camera tracker improvements - Add support for refining the camera's intrinsic parameters during a solve. Currently, refining supports only the following combinations of intrinsic parameters: f f, cx, cy f, cx, cy, k1, k2 f, k1 f, k1, k2 This is not the same as autocalibration, since the user must still make a reasonable initial guess about the focal length and other parameters, whereas true autocalibration would eliminate the need for the user specify intrinsic parameters at all. However, the solver works well with only rough guesses for the focal length, so perhaps full autocalibation is not that important. Adding support for the last two combinations, (f, k1) and (f, k1, k2) required changes to the library libmv depends on for bundle adjustment, SSBA. These changes should get ported upstream not just to libmv but to SSBA as well. - Improved the region of convergence for bundle adjustment by increasing the number of Levenberg-Marquardt iterations from 50 to 500. This way, the solver is able to crawl out of the bad local minima it gets stuck in when changing from, for example, bundling k1 and k2 to just k1 and resetting k2 to 0. - Add several new region tracker implementations. A region tracker is a libmv concept, which refers to tracking a template image pattern through frames. The impact to end users is that tracking should "just work better". I am reserving a more detailed writeup, and maybe a paper, for later. - Other libmv tweaks, such as detecting that a tracker is headed outside of the image bounds. This includes several changes made directly to the libmv extern code rather expecting to get those changes through normal libmv channels, because I, the libmv BDFL, decided it was faster to work on libmv directly in Blender, then later reverse-port the libmv changes from Blender back into libmv trunk. The interesting part is that I added a full Levenberg-Marquardt loop to the region tracking code, which should lead to a more stable solutions. I also added a hacky implementation of "Efficient Second-Order Minimization" for tracking, which works nicely. A more detailed quantitative evaluation will follow. Original patch by Keir, cleaned a bit by myself.
2011-11-14 06:41:23 +00:00
static void cameraIntrinsicsFromOptions(libmv::CameraIntrinsics *camera_intrinsics,
libmv_cameraIntrinsicsOptions *camera_intrinsics_options)
{
camera_intrinsics->SetFocalLength(camera_intrinsics_options->focal_length,
camera_intrinsics_options->focal_length);
camera_intrinsics->SetPrincipalPoint(camera_intrinsics_options->principal_point_x,
camera_intrinsics_options->principal_point_y);
camera_intrinsics->SetRadialDistortion(camera_intrinsics_options->k1,
camera_intrinsics_options->k2,
camera_intrinsics_options->k3);
camera_intrinsics->SetImageSize(camera_intrinsics_options->image_width,
camera_intrinsics_options->image_height);
}
static libmv::Tracks getNormalizedTracks(libmv::Tracks *tracks, libmv::CameraIntrinsics *camera_intrinsics)
{
libmv::vector<libmv::Marker> markers = tracks->AllMarkers();
for (int i = 0; i < markers.size(); ++i) {
camera_intrinsics->InvertIntrinsics(markers[i].x, markers[i].y,
&(markers[i].x), &(markers[i].y));
}
return libmv::Tracks(markers);
}
static void finishReconstruction(libmv::Tracks *tracks, libmv::CameraIntrinsics *camera_intrinsics,
libmv_Reconstruction *libmv_reconstruction,
reconstruct_progress_update_cb progress_update_callback,
void *callback_customdata)
{
libmv::EuclideanReconstruction *reconstruction = &libmv_reconstruction->reconstruction;
/* reprojection error calculation */
progress_update_callback(callback_customdata, 1.0, "Finishing solution");
libmv_reconstruction->tracks = *tracks;
libmv_reconstruction->error = libmv::EuclideanReprojectionError(*tracks, *reconstruction, *camera_intrinsics);
}
libmv_Reconstruction *libmv_solveReconstruction(libmv_Tracks *libmv_tracks,
libmv_cameraIntrinsicsOptions *libmv_camera_intrinsics_options,
libmv_reconstructionOptions *libmv_reconstruction_options,
reconstruct_progress_update_cb progress_update_callback,
void *callback_customdata)
Camera tracking integration =========================== Commiting camera tracking integration gsoc project into trunk. This commit includes: - Bundled version of libmv library (with some changes against official repo, re-sync with libmv repo a bit later) - New datatype ID called MovieClip which is optimized to work with movie clips (both of movie files and image sequences) and doing camera/motion tracking operations. - New editor called Clip Editor which is currently used for motion/tracking stuff only, but which can be easily extended to work with masks too. This editor supports: * Loading movie files/image sequences * Build proxies with different size for loaded movie clip, also supports building undistorted proxies to increase speed of playback in undistorted mode. * Manual lens distortion mode calibration using grid and grease pencil * Supervised 2D tracking using two different algorithms KLT and SAD. * Basic algorithm for feature detection * Camera motion solving. scene orientation - New constraints to "link" scene objects with solved motions from clip: * Follow Track (make object follow 2D motion of track with given name or parent object to reconstructed 3D position of track) * Camera Solver to make camera moving in the same way as reconstructed camera This commit NOT includes changes from tomato branch: - New nodes (they'll be commited as separated patch) - Automatic image offset guessing for image input node and image editor (need to do more tests and gather more feedback) - Code cleanup in libmv-capi. It's not so critical cleanup, just increasing readability and understanadability of code. Better to make this chaneg when Keir will finish his current patch. More details about this project can be found on this page: http://wiki.blender.org/index.php/User:Nazg-gul/GSoC-2011 Further development of small features would be done in trunk, bigger/experimental features would first be implemented in tomato branch.
2011-11-07 12:55:18 +00:00
{
libmv_Reconstruction *libmv_reconstruction = new libmv_Reconstruction();
libmv::Tracks *tracks = ((libmv::Tracks *) libmv_tracks);
Camera tracking integration =========================== Commiting camera tracking integration gsoc project into trunk. This commit includes: - Bundled version of libmv library (with some changes against official repo, re-sync with libmv repo a bit later) - New datatype ID called MovieClip which is optimized to work with movie clips (both of movie files and image sequences) and doing camera/motion tracking operations. - New editor called Clip Editor which is currently used for motion/tracking stuff only, but which can be easily extended to work with masks too. This editor supports: * Loading movie files/image sequences * Build proxies with different size for loaded movie clip, also supports building undistorted proxies to increase speed of playback in undistorted mode. * Manual lens distortion mode calibration using grid and grease pencil * Supervised 2D tracking using two different algorithms KLT and SAD. * Basic algorithm for feature detection * Camera motion solving. scene orientation - New constraints to "link" scene objects with solved motions from clip: * Follow Track (make object follow 2D motion of track with given name or parent object to reconstructed 3D position of track) * Camera Solver to make camera moving in the same way as reconstructed camera This commit NOT includes changes from tomato branch: - New nodes (they'll be commited as separated patch) - Automatic image offset guessing for image input node and image editor (need to do more tests and gather more feedback) - Code cleanup in libmv-capi. It's not so critical cleanup, just increasing readability and understanadability of code. Better to make this chaneg when Keir will finish his current patch. More details about this project can be found on this page: http://wiki.blender.org/index.php/User:Nazg-gul/GSoC-2011 Further development of small features would be done in trunk, bigger/experimental features would first be implemented in tomato branch.
2011-11-07 12:55:18 +00:00
libmv::EuclideanReconstruction *reconstruction = &libmv_reconstruction->reconstruction;
libmv::CameraIntrinsics *camera_intrinsics = &libmv_reconstruction->intrinsics;
Camera tracking integration =========================== Commiting camera tracking integration gsoc project into trunk. This commit includes: - Bundled version of libmv library (with some changes against official repo, re-sync with libmv repo a bit later) - New datatype ID called MovieClip which is optimized to work with movie clips (both of movie files and image sequences) and doing camera/motion tracking operations. - New editor called Clip Editor which is currently used for motion/tracking stuff only, but which can be easily extended to work with masks too. This editor supports: * Loading movie files/image sequences * Build proxies with different size for loaded movie clip, also supports building undistorted proxies to increase speed of playback in undistorted mode. * Manual lens distortion mode calibration using grid and grease pencil * Supervised 2D tracking using two different algorithms KLT and SAD. * Basic algorithm for feature detection * Camera motion solving. scene orientation - New constraints to "link" scene objects with solved motions from clip: * Follow Track (make object follow 2D motion of track with given name or parent object to reconstructed 3D position of track) * Camera Solver to make camera moving in the same way as reconstructed camera This commit NOT includes changes from tomato branch: - New nodes (they'll be commited as separated patch) - Automatic image offset guessing for image input node and image editor (need to do more tests and gather more feedback) - Code cleanup in libmv-capi. It's not so critical cleanup, just increasing readability and understanadability of code. Better to make this chaneg when Keir will finish his current patch. More details about this project can be found on this page: http://wiki.blender.org/index.php/User:Nazg-gul/GSoC-2011 Further development of small features would be done in trunk, bigger/experimental features would first be implemented in tomato branch.
2011-11-07 12:55:18 +00:00
ReconstructUpdateCallback update_callback =
ReconstructUpdateCallback(progress_update_callback, callback_customdata);
cameraIntrinsicsFromOptions(camera_intrinsics, libmv_camera_intrinsics_options);
Camera tracking integration =========================== Commiting camera tracking integration gsoc project into trunk. This commit includes: - Bundled version of libmv library (with some changes against official repo, re-sync with libmv repo a bit later) - New datatype ID called MovieClip which is optimized to work with movie clips (both of movie files and image sequences) and doing camera/motion tracking operations. - New editor called Clip Editor which is currently used for motion/tracking stuff only, but which can be easily extended to work with masks too. This editor supports: * Loading movie files/image sequences * Build proxies with different size for loaded movie clip, also supports building undistorted proxies to increase speed of playback in undistorted mode. * Manual lens distortion mode calibration using grid and grease pencil * Supervised 2D tracking using two different algorithms KLT and SAD. * Basic algorithm for feature detection * Camera motion solving. scene orientation - New constraints to "link" scene objects with solved motions from clip: * Follow Track (make object follow 2D motion of track with given name or parent object to reconstructed 3D position of track) * Camera Solver to make camera moving in the same way as reconstructed camera This commit NOT includes changes from tomato branch: - New nodes (they'll be commited as separated patch) - Automatic image offset guessing for image input node and image editor (need to do more tests and gather more feedback) - Code cleanup in libmv-capi. It's not so critical cleanup, just increasing readability and understanadability of code. Better to make this chaneg when Keir will finish his current patch. More details about this project can be found on this page: http://wiki.blender.org/index.php/User:Nazg-gul/GSoC-2011 Further development of small features would be done in trunk, bigger/experimental features would first be implemented in tomato branch.
2011-11-07 12:55:18 +00:00
/* Invert the camera intrinsics */
libmv::Tracks normalized_tracks = getNormalizedTracks(tracks, camera_intrinsics);
/* actual reconstruction */
libmv::ReconstructionOptions reconstruction_options;
reconstruction_options.success_threshold = libmv_reconstruction_options->success_threshold;
reconstruction_options.use_fallback_reconstruction = libmv_reconstruction_options->use_fallback_reconstruction;
Camera tracking integration =========================== Commiting camera tracking integration gsoc project into trunk. This commit includes: - Bundled version of libmv library (with some changes against official repo, re-sync with libmv repo a bit later) - New datatype ID called MovieClip which is optimized to work with movie clips (both of movie files and image sequences) and doing camera/motion tracking operations. - New editor called Clip Editor which is currently used for motion/tracking stuff only, but which can be easily extended to work with masks too. This editor supports: * Loading movie files/image sequences * Build proxies with different size for loaded movie clip, also supports building undistorted proxies to increase speed of playback in undistorted mode. * Manual lens distortion mode calibration using grid and grease pencil * Supervised 2D tracking using two different algorithms KLT and SAD. * Basic algorithm for feature detection * Camera motion solving. scene orientation - New constraints to "link" scene objects with solved motions from clip: * Follow Track (make object follow 2D motion of track with given name or parent object to reconstructed 3D position of track) * Camera Solver to make camera moving in the same way as reconstructed camera This commit NOT includes changes from tomato branch: - New nodes (they'll be commited as separated patch) - Automatic image offset guessing for image input node and image editor (need to do more tests and gather more feedback) - Code cleanup in libmv-capi. It's not so critical cleanup, just increasing readability and understanadability of code. Better to make this chaneg when Keir will finish his current patch. More details about this project can be found on this page: http://wiki.blender.org/index.php/User:Nazg-gul/GSoC-2011 Further development of small features would be done in trunk, bigger/experimental features would first be implemented in tomato branch.
2011-11-07 12:55:18 +00:00
int keyframe1 = libmv_reconstruction_options->keyframe1,
keyframe2 = libmv_reconstruction_options->keyframe2;
Camera tracking integration =========================== Commiting camera tracking integration gsoc project into trunk. This commit includes: - Bundled version of libmv library (with some changes against official repo, re-sync with libmv repo a bit later) - New datatype ID called MovieClip which is optimized to work with movie clips (both of movie files and image sequences) and doing camera/motion tracking operations. - New editor called Clip Editor which is currently used for motion/tracking stuff only, but which can be easily extended to work with masks too. This editor supports: * Loading movie files/image sequences * Build proxies with different size for loaded movie clip, also supports building undistorted proxies to increase speed of playback in undistorted mode. * Manual lens distortion mode calibration using grid and grease pencil * Supervised 2D tracking using two different algorithms KLT and SAD. * Basic algorithm for feature detection * Camera motion solving. scene orientation - New constraints to "link" scene objects with solved motions from clip: * Follow Track (make object follow 2D motion of track with given name or parent object to reconstructed 3D position of track) * Camera Solver to make camera moving in the same way as reconstructed camera This commit NOT includes changes from tomato branch: - New nodes (they'll be commited as separated patch) - Automatic image offset guessing for image input node and image editor (need to do more tests and gather more feedback) - Code cleanup in libmv-capi. It's not so critical cleanup, just increasing readability and understanadability of code. Better to make this chaneg when Keir will finish his current patch. More details about this project can be found on this page: http://wiki.blender.org/index.php/User:Nazg-gul/GSoC-2011 Further development of small features would be done in trunk, bigger/experimental features would first be implemented in tomato branch.
2011-11-07 12:55:18 +00:00
LG << "frames to init from: " << keyframe1 << " " << keyframe2;
Camera tracking integration =========================== Commiting camera tracking integration gsoc project into trunk. This commit includes: - Bundled version of libmv library (with some changes against official repo, re-sync with libmv repo a bit later) - New datatype ID called MovieClip which is optimized to work with movie clips (both of movie files and image sequences) and doing camera/motion tracking operations. - New editor called Clip Editor which is currently used for motion/tracking stuff only, but which can be easily extended to work with masks too. This editor supports: * Loading movie files/image sequences * Build proxies with different size for loaded movie clip, also supports building undistorted proxies to increase speed of playback in undistorted mode. * Manual lens distortion mode calibration using grid and grease pencil * Supervised 2D tracking using two different algorithms KLT and SAD. * Basic algorithm for feature detection * Camera motion solving. scene orientation - New constraints to "link" scene objects with solved motions from clip: * Follow Track (make object follow 2D motion of track with given name or parent object to reconstructed 3D position of track) * Camera Solver to make camera moving in the same way as reconstructed camera This commit NOT includes changes from tomato branch: - New nodes (they'll be commited as separated patch) - Automatic image offset guessing for image input node and image editor (need to do more tests and gather more feedback) - Code cleanup in libmv-capi. It's not so critical cleanup, just increasing readability and understanadability of code. Better to make this chaneg when Keir will finish his current patch. More details about this project can be found on this page: http://wiki.blender.org/index.php/User:Nazg-gul/GSoC-2011 Further development of small features would be done in trunk, bigger/experimental features would first be implemented in tomato branch.
2011-11-07 12:55:18 +00:00
libmv::vector<libmv::Marker> keyframe_markers =
normalized_tracks.MarkersForTracksInBothImages(keyframe1, keyframe2);
LG << "number of markers for init: " << keyframe_markers.size();
update_callback.invoke(0, "Initial reconstruction");
Camera tracking integration =========================== Commiting camera tracking integration gsoc project into trunk. This commit includes: - Bundled version of libmv library (with some changes against official repo, re-sync with libmv repo a bit later) - New datatype ID called MovieClip which is optimized to work with movie clips (both of movie files and image sequences) and doing camera/motion tracking operations. - New editor called Clip Editor which is currently used for motion/tracking stuff only, but which can be easily extended to work with masks too. This editor supports: * Loading movie files/image sequences * Build proxies with different size for loaded movie clip, also supports building undistorted proxies to increase speed of playback in undistorted mode. * Manual lens distortion mode calibration using grid and grease pencil * Supervised 2D tracking using two different algorithms KLT and SAD. * Basic algorithm for feature detection * Camera motion solving. scene orientation - New constraints to "link" scene objects with solved motions from clip: * Follow Track (make object follow 2D motion of track with given name or parent object to reconstructed 3D position of track) * Camera Solver to make camera moving in the same way as reconstructed camera This commit NOT includes changes from tomato branch: - New nodes (they'll be commited as separated patch) - Automatic image offset guessing for image input node and image editor (need to do more tests and gather more feedback) - Code cleanup in libmv-capi. It's not so critical cleanup, just increasing readability and understanadability of code. Better to make this chaneg when Keir will finish his current patch. More details about this project can be found on this page: http://wiki.blender.org/index.php/User:Nazg-gul/GSoC-2011 Further development of small features would be done in trunk, bigger/experimental features would first be implemented in tomato branch.
2011-11-07 12:55:18 +00:00
libmv::EuclideanReconstructTwoFrames(keyframe_markers, reconstruction);
libmv::EuclideanBundle(normalized_tracks, reconstruction);
libmv::EuclideanCompleteReconstruction(reconstruction_options, normalized_tracks,
reconstruction, &update_callback);
Camera tracking integration =========================== Commiting camera tracking integration gsoc project into trunk. This commit includes: - Bundled version of libmv library (with some changes against official repo, re-sync with libmv repo a bit later) - New datatype ID called MovieClip which is optimized to work with movie clips (both of movie files and image sequences) and doing camera/motion tracking operations. - New editor called Clip Editor which is currently used for motion/tracking stuff only, but which can be easily extended to work with masks too. This editor supports: * Loading movie files/image sequences * Build proxies with different size for loaded movie clip, also supports building undistorted proxies to increase speed of playback in undistorted mode. * Manual lens distortion mode calibration using grid and grease pencil * Supervised 2D tracking using two different algorithms KLT and SAD. * Basic algorithm for feature detection * Camera motion solving. scene orientation - New constraints to "link" scene objects with solved motions from clip: * Follow Track (make object follow 2D motion of track with given name or parent object to reconstructed 3D position of track) * Camera Solver to make camera moving in the same way as reconstructed camera This commit NOT includes changes from tomato branch: - New nodes (they'll be commited as separated patch) - Automatic image offset guessing for image input node and image editor (need to do more tests and gather more feedback) - Code cleanup in libmv-capi. It's not so critical cleanup, just increasing readability and understanadability of code. Better to make this chaneg when Keir will finish his current patch. More details about this project can be found on this page: http://wiki.blender.org/index.php/User:Nazg-gul/GSoC-2011 Further development of small features would be done in trunk, bigger/experimental features would first be implemented in tomato branch.
2011-11-07 12:55:18 +00:00
/* refinement */
if (libmv_reconstruction_options->refine_intrinsics) {
libmv_solveRefineIntrinsics((libmv::Tracks *)tracks, camera_intrinsics, reconstruction,
libmv_reconstruction_options->refine_intrinsics,
progress_update_callback, callback_customdata);
}
/* finish reconstruction */
finishReconstruction(tracks, camera_intrinsics, libmv_reconstruction,
progress_update_callback, callback_customdata);
return (libmv_Reconstruction *)libmv_reconstruction;
}
struct libmv_Reconstruction *libmv_solveModal(struct libmv_Tracks *libmv_tracks,
libmv_cameraIntrinsicsOptions *libmv_camera_intrinsics_options,
libmv_reconstructionOptions *libmv_reconstruction_options,
reconstruct_progress_update_cb progress_update_callback,
void *callback_customdata)
{
libmv_Reconstruction *libmv_reconstruction = new libmv_Reconstruction();
libmv::Tracks *tracks = ((libmv::Tracks *) libmv_tracks);
libmv::EuclideanReconstruction *reconstruction = &libmv_reconstruction->reconstruction;
libmv::CameraIntrinsics *camera_intrinsics = &libmv_reconstruction->intrinsics;
ReconstructUpdateCallback update_callback =
ReconstructUpdateCallback(progress_update_callback, callback_customdata);
cameraIntrinsicsFromOptions(camera_intrinsics, libmv_camera_intrinsics_options);
/* Invert the camera intrinsics. */
libmv::Tracks normalized_tracks = getNormalizedTracks(tracks, camera_intrinsics);
/* Actual reconstruction. */
libmv::ModalSolver(normalized_tracks, reconstruction, &update_callback);
libmv::CameraIntrinsics empty_intrinsics;
libmv::EuclideanBundleCommonIntrinsics(normalized_tracks,
libmv::BUNDLE_NO_INTRINSICS,
reconstruction,
&empty_intrinsics,
libmv::BUNDLE_NO_TRANSLATION);
/* Refinement. */
if (libmv_reconstruction_options->refine_intrinsics) {
libmv_solveRefineIntrinsics((libmv::Tracks *)tracks, camera_intrinsics, reconstruction,
libmv_reconstruction_options->refine_intrinsics,
progress_update_callback, callback_customdata,
libmv::BUNDLE_NO_TRANSLATION);
}
/* Finish reconstruction. */
finishReconstruction(tracks, camera_intrinsics, libmv_reconstruction,
progress_update_callback, callback_customdata);
Camera tracking integration =========================== Commiting camera tracking integration gsoc project into trunk. This commit includes: - Bundled version of libmv library (with some changes against official repo, re-sync with libmv repo a bit later) - New datatype ID called MovieClip which is optimized to work with movie clips (both of movie files and image sequences) and doing camera/motion tracking operations. - New editor called Clip Editor which is currently used for motion/tracking stuff only, but which can be easily extended to work with masks too. This editor supports: * Loading movie files/image sequences * Build proxies with different size for loaded movie clip, also supports building undistorted proxies to increase speed of playback in undistorted mode. * Manual lens distortion mode calibration using grid and grease pencil * Supervised 2D tracking using two different algorithms KLT and SAD. * Basic algorithm for feature detection * Camera motion solving. scene orientation - New constraints to "link" scene objects with solved motions from clip: * Follow Track (make object follow 2D motion of track with given name or parent object to reconstructed 3D position of track) * Camera Solver to make camera moving in the same way as reconstructed camera This commit NOT includes changes from tomato branch: - New nodes (they'll be commited as separated patch) - Automatic image offset guessing for image input node and image editor (need to do more tests and gather more feedback) - Code cleanup in libmv-capi. It's not so critical cleanup, just increasing readability and understanadability of code. Better to make this chaneg when Keir will finish his current patch. More details about this project can be found on this page: http://wiki.blender.org/index.php/User:Nazg-gul/GSoC-2011 Further development of small features would be done in trunk, bigger/experimental features would first be implemented in tomato branch.
2011-11-07 12:55:18 +00:00
return (libmv_Reconstruction *)libmv_reconstruction;
}
int libmv_reporojectionPointForTrack(libmv_Reconstruction *libmv_reconstruction, int track, double pos[3])
{
libmv::EuclideanReconstruction *reconstruction = &libmv_reconstruction->reconstruction;
libmv::EuclideanPoint *point = reconstruction->PointForTrack(track);
if(point) {
pos[0] = point->X[0];
pos[1] = point->X[2];
pos[2] = point->X[1];
return 1;
}
return 0;
}
static libmv::Marker ProjectMarker(const libmv::EuclideanPoint &point, const libmv::EuclideanCamera &camera,
const libmv::CameraIntrinsics &intrinsics) {
libmv::Vec3 projected = camera.R * point.X + camera.t;
projected /= projected(2);
libmv::Marker reprojected_marker;
intrinsics.ApplyIntrinsics(projected(0), projected(1), &reprojected_marker.x, &reprojected_marker.y);
reprojected_marker.image = camera.image;
reprojected_marker.track = point.track;
return reprojected_marker;
}
double libmv_reporojectionErrorForTrack(libmv_Reconstruction *libmv_reconstruction, int track)
{
libmv::EuclideanReconstruction *reconstruction = &libmv_reconstruction->reconstruction;
libmv::CameraIntrinsics *intrinsics = &libmv_reconstruction->intrinsics;
2011-12-05 07:56:31 +00:00
libmv::vector<libmv::Marker> markers = libmv_reconstruction->tracks.MarkersForTrack(track);
Camera tracking integration =========================== Commiting camera tracking integration gsoc project into trunk. This commit includes: - Bundled version of libmv library (with some changes against official repo, re-sync with libmv repo a bit later) - New datatype ID called MovieClip which is optimized to work with movie clips (both of movie files and image sequences) and doing camera/motion tracking operations. - New editor called Clip Editor which is currently used for motion/tracking stuff only, but which can be easily extended to work with masks too. This editor supports: * Loading movie files/image sequences * Build proxies with different size for loaded movie clip, also supports building undistorted proxies to increase speed of playback in undistorted mode. * Manual lens distortion mode calibration using grid and grease pencil * Supervised 2D tracking using two different algorithms KLT and SAD. * Basic algorithm for feature detection * Camera motion solving. scene orientation - New constraints to "link" scene objects with solved motions from clip: * Follow Track (make object follow 2D motion of track with given name or parent object to reconstructed 3D position of track) * Camera Solver to make camera moving in the same way as reconstructed camera This commit NOT includes changes from tomato branch: - New nodes (they'll be commited as separated patch) - Automatic image offset guessing for image input node and image editor (need to do more tests and gather more feedback) - Code cleanup in libmv-capi. It's not so critical cleanup, just increasing readability and understanadability of code. Better to make this chaneg when Keir will finish his current patch. More details about this project can be found on this page: http://wiki.blender.org/index.php/User:Nazg-gul/GSoC-2011 Further development of small features would be done in trunk, bigger/experimental features would first be implemented in tomato branch.
2011-11-07 12:55:18 +00:00
int num_reprojected = 0;
double total_error = 0.0;
for (int i = 0; i < markers.size(); ++i) {
const libmv::EuclideanCamera *camera = reconstruction->CameraForImage(markers[i].image);
const libmv::EuclideanPoint *point = reconstruction->PointForTrack(markers[i].track);
if (!camera || !point) {
continue;
}
num_reprojected++;
libmv::Marker reprojected_marker = ProjectMarker(*point, *camera, *intrinsics);
double ex = reprojected_marker.x - markers[i].x;
double ey = reprojected_marker.y - markers[i].y;
total_error += sqrt(ex*ex + ey*ey);
}
return total_error / num_reprojected;
}
double libmv_reporojectionErrorForImage(libmv_Reconstruction *libmv_reconstruction, int image)
{
libmv::EuclideanReconstruction *reconstruction = &libmv_reconstruction->reconstruction;
libmv::CameraIntrinsics *intrinsics = &libmv_reconstruction->intrinsics;
libmv::vector<libmv::Marker> markers = libmv_reconstruction->tracks.MarkersInImage(image);
const libmv::EuclideanCamera *camera = reconstruction->CameraForImage(image);
int num_reprojected = 0;
double total_error = 0.0;
if (!camera)
return 0;
for (int i = 0; i < markers.size(); ++i) {
const libmv::EuclideanPoint *point = reconstruction->PointForTrack(markers[i].track);
if (!point) {
continue;
}
num_reprojected++;
libmv::Marker reprojected_marker = ProjectMarker(*point, *camera, *intrinsics);
double ex = reprojected_marker.x - markers[i].x;
double ey = reprojected_marker.y - markers[i].y;
total_error += sqrt(ex*ex + ey*ey);
}
return total_error / num_reprojected;
}
int libmv_reporojectionCameraForImage(libmv_Reconstruction *libmv_reconstruction, int image, double mat[4][4])
{
libmv::EuclideanReconstruction *reconstruction = &libmv_reconstruction->reconstruction;
libmv::EuclideanCamera *camera = reconstruction->CameraForImage(image);
if(camera) {
for (int j = 0; j < 3; ++j) {
for (int k = 0; k < 3; ++k) {
int l = k;
if (k == 1) l = 2;
else if (k == 2) l = 1;
if (j == 2) mat[j][l] = -camera->R(j,k);
else mat[j][l] = camera->R(j,k);
}
mat[j][3]= 0.0;
}
libmv::Vec3 optical_center = -camera->R.transpose() * camera->t;
mat[3][0] = optical_center(0);
mat[3][1] = optical_center(2);
mat[3][2] = optical_center(1);
mat[3][3]= 1.0;
return 1;
}
return 0;
}
double libmv_reprojectionError(libmv_Reconstruction *libmv_reconstruction)
{
return libmv_reconstruction->error;
}
void libmv_destroyReconstruction(libmv_Reconstruction *libmv_reconstruction)
{
delete libmv_reconstruction;
}
/* ************ feature detector ************ */
struct libmv_Features *libmv_detectFeaturesFAST(unsigned char *data, int width, int height, int stride,
int margin, int min_trackness, int min_distance)
{
libmv::Feature *features = NULL;
std::vector<libmv::Feature> v;
libmv_Features *libmv_features = new libmv_Features();
int i= 0, count;
if(margin) {
data += margin*stride+margin;
width -= 2*margin;
height -= 2*margin;
}
v = libmv::DetectFAST(data, width, height, stride, min_trackness, min_distance);
count = v.size();
if(count) {
features= new libmv::Feature[count];
for(std::vector<libmv::Feature>::iterator it = v.begin(); it != v.end(); it++) {
features[i++]= *it;
}
}
libmv_features->features = features;
libmv_features->count = count;
libmv_features->margin = margin;
return (libmv_Features *)libmv_features;
}
struct libmv_Features *libmv_detectFeaturesMORAVEC(unsigned char *data, int width, int height, int stride,
int margin, int count, int min_distance)
{
libmv::Feature *features = NULL;
libmv_Features *libmv_features = new libmv_Features;
if(count) {
if(margin) {
data += margin*stride+margin;
width -= 2*margin;
height -= 2*margin;
}
features = new libmv::Feature[count];
libmv::DetectMORAVEC(data, stride, width, height, features, &count, min_distance, NULL);
}
libmv_features->count = count;
libmv_features->margin = margin;
libmv_features->features = features;
return libmv_features;
}
int libmv_countFeatures(struct libmv_Features *libmv_features)
{
return libmv_features->count;
}
void libmv_getFeature(struct libmv_Features *libmv_features, int number, double *x, double *y, double *score, double *size)
{
libmv::Feature feature= libmv_features->features[number];
*x = feature.x + libmv_features->margin;
*y = feature.y + libmv_features->margin;
*score = feature.score;
*size = feature.size;
}
void libmv_destroyFeatures(struct libmv_Features *libmv_features)
{
if(libmv_features->features)
delete [] libmv_features->features;
delete libmv_features;
}
/* ************ camera intrinsics ************ */
Assorted camera tracker improvements - Add support for refining the camera's intrinsic parameters during a solve. Currently, refining supports only the following combinations of intrinsic parameters: f f, cx, cy f, cx, cy, k1, k2 f, k1 f, k1, k2 This is not the same as autocalibration, since the user must still make a reasonable initial guess about the focal length and other parameters, whereas true autocalibration would eliminate the need for the user specify intrinsic parameters at all. However, the solver works well with only rough guesses for the focal length, so perhaps full autocalibation is not that important. Adding support for the last two combinations, (f, k1) and (f, k1, k2) required changes to the library libmv depends on for bundle adjustment, SSBA. These changes should get ported upstream not just to libmv but to SSBA as well. - Improved the region of convergence for bundle adjustment by increasing the number of Levenberg-Marquardt iterations from 50 to 500. This way, the solver is able to crawl out of the bad local minima it gets stuck in when changing from, for example, bundling k1 and k2 to just k1 and resetting k2 to 0. - Add several new region tracker implementations. A region tracker is a libmv concept, which refers to tracking a template image pattern through frames. The impact to end users is that tracking should "just work better". I am reserving a more detailed writeup, and maybe a paper, for later. - Other libmv tweaks, such as detecting that a tracker is headed outside of the image bounds. This includes several changes made directly to the libmv extern code rather expecting to get those changes through normal libmv channels, because I, the libmv BDFL, decided it was faster to work on libmv directly in Blender, then later reverse-port the libmv changes from Blender back into libmv trunk. The interesting part is that I added a full Levenberg-Marquardt loop to the region tracking code, which should lead to a more stable solutions. I also added a hacky implementation of "Efficient Second-Order Minimization" for tracking, which works nicely. A more detailed quantitative evaluation will follow. Original patch by Keir, cleaned a bit by myself.
2011-11-14 06:41:23 +00:00
struct libmv_CameraIntrinsics *libmv_ReconstructionExtractIntrinsics(struct libmv_Reconstruction *libmv_Reconstruction) {
2011-12-05 07:56:31 +00:00
return (struct libmv_CameraIntrinsics *)&libmv_Reconstruction->intrinsics;
Assorted camera tracker improvements - Add support for refining the camera's intrinsic parameters during a solve. Currently, refining supports only the following combinations of intrinsic parameters: f f, cx, cy f, cx, cy, k1, k2 f, k1 f, k1, k2 This is not the same as autocalibration, since the user must still make a reasonable initial guess about the focal length and other parameters, whereas true autocalibration would eliminate the need for the user specify intrinsic parameters at all. However, the solver works well with only rough guesses for the focal length, so perhaps full autocalibation is not that important. Adding support for the last two combinations, (f, k1) and (f, k1, k2) required changes to the library libmv depends on for bundle adjustment, SSBA. These changes should get ported upstream not just to libmv but to SSBA as well. - Improved the region of convergence for bundle adjustment by increasing the number of Levenberg-Marquardt iterations from 50 to 500. This way, the solver is able to crawl out of the bad local minima it gets stuck in when changing from, for example, bundling k1 and k2 to just k1 and resetting k2 to 0. - Add several new region tracker implementations. A region tracker is a libmv concept, which refers to tracking a template image pattern through frames. The impact to end users is that tracking should "just work better". I am reserving a more detailed writeup, and maybe a paper, for later. - Other libmv tweaks, such as detecting that a tracker is headed outside of the image bounds. This includes several changes made directly to the libmv extern code rather expecting to get those changes through normal libmv channels, because I, the libmv BDFL, decided it was faster to work on libmv directly in Blender, then later reverse-port the libmv changes from Blender back into libmv trunk. The interesting part is that I added a full Levenberg-Marquardt loop to the region tracking code, which should lead to a more stable solutions. I also added a hacky implementation of "Efficient Second-Order Minimization" for tracking, which works nicely. A more detailed quantitative evaluation will follow. Original patch by Keir, cleaned a bit by myself.
2011-11-14 06:41:23 +00:00
}
struct libmv_CameraIntrinsics *libmv_CameraIntrinsicsNewEmpty(void)
{
libmv::CameraIntrinsics *camera_intrinsics = new libmv::CameraIntrinsics();
return (struct libmv_CameraIntrinsics *) camera_intrinsics;
}
struct libmv_CameraIntrinsics *libmv_CameraIntrinsicsNew(libmv_cameraIntrinsicsOptions *libmv_camera_intrinsics_options)
Camera tracking integration =========================== Commiting camera tracking integration gsoc project into trunk. This commit includes: - Bundled version of libmv library (with some changes against official repo, re-sync with libmv repo a bit later) - New datatype ID called MovieClip which is optimized to work with movie clips (both of movie files and image sequences) and doing camera/motion tracking operations. - New editor called Clip Editor which is currently used for motion/tracking stuff only, but which can be easily extended to work with masks too. This editor supports: * Loading movie files/image sequences * Build proxies with different size for loaded movie clip, also supports building undistorted proxies to increase speed of playback in undistorted mode. * Manual lens distortion mode calibration using grid and grease pencil * Supervised 2D tracking using two different algorithms KLT and SAD. * Basic algorithm for feature detection * Camera motion solving. scene orientation - New constraints to "link" scene objects with solved motions from clip: * Follow Track (make object follow 2D motion of track with given name or parent object to reconstructed 3D position of track) * Camera Solver to make camera moving in the same way as reconstructed camera This commit NOT includes changes from tomato branch: - New nodes (they'll be commited as separated patch) - Automatic image offset guessing for image input node and image editor (need to do more tests and gather more feedback) - Code cleanup in libmv-capi. It's not so critical cleanup, just increasing readability and understanadability of code. Better to make this chaneg when Keir will finish his current patch. More details about this project can be found on this page: http://wiki.blender.org/index.php/User:Nazg-gul/GSoC-2011 Further development of small features would be done in trunk, bigger/experimental features would first be implemented in tomato branch.
2011-11-07 12:55:18 +00:00
{
libmv::CameraIntrinsics *camera_intrinsics = new libmv::CameraIntrinsics();
Camera tracking integration =========================== Commiting camera tracking integration gsoc project into trunk. This commit includes: - Bundled version of libmv library (with some changes against official repo, re-sync with libmv repo a bit later) - New datatype ID called MovieClip which is optimized to work with movie clips (both of movie files and image sequences) and doing camera/motion tracking operations. - New editor called Clip Editor which is currently used for motion/tracking stuff only, but which can be easily extended to work with masks too. This editor supports: * Loading movie files/image sequences * Build proxies with different size for loaded movie clip, also supports building undistorted proxies to increase speed of playback in undistorted mode. * Manual lens distortion mode calibration using grid and grease pencil * Supervised 2D tracking using two different algorithms KLT and SAD. * Basic algorithm for feature detection * Camera motion solving. scene orientation - New constraints to "link" scene objects with solved motions from clip: * Follow Track (make object follow 2D motion of track with given name or parent object to reconstructed 3D position of track) * Camera Solver to make camera moving in the same way as reconstructed camera This commit NOT includes changes from tomato branch: - New nodes (they'll be commited as separated patch) - Automatic image offset guessing for image input node and image editor (need to do more tests and gather more feedback) - Code cleanup in libmv-capi. It's not so critical cleanup, just increasing readability and understanadability of code. Better to make this chaneg when Keir will finish his current patch. More details about this project can be found on this page: http://wiki.blender.org/index.php/User:Nazg-gul/GSoC-2011 Further development of small features would be done in trunk, bigger/experimental features would first be implemented in tomato branch.
2011-11-07 12:55:18 +00:00
cameraIntrinsicsFromOptions(camera_intrinsics, libmv_camera_intrinsics_options);
Camera tracking integration =========================== Commiting camera tracking integration gsoc project into trunk. This commit includes: - Bundled version of libmv library (with some changes against official repo, re-sync with libmv repo a bit later) - New datatype ID called MovieClip which is optimized to work with movie clips (both of movie files and image sequences) and doing camera/motion tracking operations. - New editor called Clip Editor which is currently used for motion/tracking stuff only, but which can be easily extended to work with masks too. This editor supports: * Loading movie files/image sequences * Build proxies with different size for loaded movie clip, also supports building undistorted proxies to increase speed of playback in undistorted mode. * Manual lens distortion mode calibration using grid and grease pencil * Supervised 2D tracking using two different algorithms KLT and SAD. * Basic algorithm for feature detection * Camera motion solving. scene orientation - New constraints to "link" scene objects with solved motions from clip: * Follow Track (make object follow 2D motion of track with given name or parent object to reconstructed 3D position of track) * Camera Solver to make camera moving in the same way as reconstructed camera This commit NOT includes changes from tomato branch: - New nodes (they'll be commited as separated patch) - Automatic image offset guessing for image input node and image editor (need to do more tests and gather more feedback) - Code cleanup in libmv-capi. It's not so critical cleanup, just increasing readability and understanadability of code. Better to make this chaneg when Keir will finish his current patch. More details about this project can be found on this page: http://wiki.blender.org/index.php/User:Nazg-gul/GSoC-2011 Further development of small features would be done in trunk, bigger/experimental features would first be implemented in tomato branch.
2011-11-07 12:55:18 +00:00
return (struct libmv_CameraIntrinsics *) camera_intrinsics;
Camera tracking integration =========================== Commiting camera tracking integration gsoc project into trunk. This commit includes: - Bundled version of libmv library (with some changes against official repo, re-sync with libmv repo a bit later) - New datatype ID called MovieClip which is optimized to work with movie clips (both of movie files and image sequences) and doing camera/motion tracking operations. - New editor called Clip Editor which is currently used for motion/tracking stuff only, but which can be easily extended to work with masks too. This editor supports: * Loading movie files/image sequences * Build proxies with different size for loaded movie clip, also supports building undistorted proxies to increase speed of playback in undistorted mode. * Manual lens distortion mode calibration using grid and grease pencil * Supervised 2D tracking using two different algorithms KLT and SAD. * Basic algorithm for feature detection * Camera motion solving. scene orientation - New constraints to "link" scene objects with solved motions from clip: * Follow Track (make object follow 2D motion of track with given name or parent object to reconstructed 3D position of track) * Camera Solver to make camera moving in the same way as reconstructed camera This commit NOT includes changes from tomato branch: - New nodes (they'll be commited as separated patch) - Automatic image offset guessing for image input node and image editor (need to do more tests and gather more feedback) - Code cleanup in libmv-capi. It's not so critical cleanup, just increasing readability and understanadability of code. Better to make this chaneg when Keir will finish his current patch. More details about this project can be found on this page: http://wiki.blender.org/index.php/User:Nazg-gul/GSoC-2011 Further development of small features would be done in trunk, bigger/experimental features would first be implemented in tomato branch.
2011-11-07 12:55:18 +00:00
}
struct libmv_CameraIntrinsics *libmv_CameraIntrinsicsCopy(struct libmv_CameraIntrinsics *libmvIntrinsics)
{
libmv::CameraIntrinsics *orig_intrinsics = (libmv::CameraIntrinsics *) libmvIntrinsics;
libmv::CameraIntrinsics *new_intrinsics= new libmv::CameraIntrinsics(*orig_intrinsics);
return (struct libmv_CameraIntrinsics *) new_intrinsics;
}
void libmv_CameraIntrinsicsDestroy(struct libmv_CameraIntrinsics *libmvIntrinsics)
{
libmv::CameraIntrinsics *intrinsics = (libmv::CameraIntrinsics *) libmvIntrinsics;
delete intrinsics;
}
void libmv_CameraIntrinsicsUpdate(struct libmv_CameraIntrinsics *libmv_intrinsics,
libmv_cameraIntrinsicsOptions *libmv_camera_intrinsics_options)
Camera tracking integration =========================== Commiting camera tracking integration gsoc project into trunk. This commit includes: - Bundled version of libmv library (with some changes against official repo, re-sync with libmv repo a bit later) - New datatype ID called MovieClip which is optimized to work with movie clips (both of movie files and image sequences) and doing camera/motion tracking operations. - New editor called Clip Editor which is currently used for motion/tracking stuff only, but which can be easily extended to work with masks too. This editor supports: * Loading movie files/image sequences * Build proxies with different size for loaded movie clip, also supports building undistorted proxies to increase speed of playback in undistorted mode. * Manual lens distortion mode calibration using grid and grease pencil * Supervised 2D tracking using two different algorithms KLT and SAD. * Basic algorithm for feature detection * Camera motion solving. scene orientation - New constraints to "link" scene objects with solved motions from clip: * Follow Track (make object follow 2D motion of track with given name or parent object to reconstructed 3D position of track) * Camera Solver to make camera moving in the same way as reconstructed camera This commit NOT includes changes from tomato branch: - New nodes (they'll be commited as separated patch) - Automatic image offset guessing for image input node and image editor (need to do more tests and gather more feedback) - Code cleanup in libmv-capi. It's not so critical cleanup, just increasing readability and understanadability of code. Better to make this chaneg when Keir will finish his current patch. More details about this project can be found on this page: http://wiki.blender.org/index.php/User:Nazg-gul/GSoC-2011 Further development of small features would be done in trunk, bigger/experimental features would first be implemented in tomato branch.
2011-11-07 12:55:18 +00:00
{
libmv::CameraIntrinsics *camera_intrinsics = (libmv::CameraIntrinsics *) libmv_intrinsics;
Camera tracking integration =========================== Commiting camera tracking integration gsoc project into trunk. This commit includes: - Bundled version of libmv library (with some changes against official repo, re-sync with libmv repo a bit later) - New datatype ID called MovieClip which is optimized to work with movie clips (both of movie files and image sequences) and doing camera/motion tracking operations. - New editor called Clip Editor which is currently used for motion/tracking stuff only, but which can be easily extended to work with masks too. This editor supports: * Loading movie files/image sequences * Build proxies with different size for loaded movie clip, also supports building undistorted proxies to increase speed of playback in undistorted mode. * Manual lens distortion mode calibration using grid and grease pencil * Supervised 2D tracking using two different algorithms KLT and SAD. * Basic algorithm for feature detection * Camera motion solving. scene orientation - New constraints to "link" scene objects with solved motions from clip: * Follow Track (make object follow 2D motion of track with given name or parent object to reconstructed 3D position of track) * Camera Solver to make camera moving in the same way as reconstructed camera This commit NOT includes changes from tomato branch: - New nodes (they'll be commited as separated patch) - Automatic image offset guessing for image input node and image editor (need to do more tests and gather more feedback) - Code cleanup in libmv-capi. It's not so critical cleanup, just increasing readability and understanadability of code. Better to make this chaneg when Keir will finish his current patch. More details about this project can be found on this page: http://wiki.blender.org/index.php/User:Nazg-gul/GSoC-2011 Further development of small features would be done in trunk, bigger/experimental features would first be implemented in tomato branch.
2011-11-07 12:55:18 +00:00
double focal_length = libmv_camera_intrinsics_options->focal_length;
double principal_x = libmv_camera_intrinsics_options->principal_point_x;
double principal_y = libmv_camera_intrinsics_options->principal_point_y;
double k1 = libmv_camera_intrinsics_options->k1;
double k2 = libmv_camera_intrinsics_options->k2;
double k3 = libmv_camera_intrinsics_options->k3;
int image_width = libmv_camera_intrinsics_options->image_width;
int image_height = libmv_camera_intrinsics_options->image_height;
Camera tracking integration =========================== Commiting camera tracking integration gsoc project into trunk. This commit includes: - Bundled version of libmv library (with some changes against official repo, re-sync with libmv repo a bit later) - New datatype ID called MovieClip which is optimized to work with movie clips (both of movie files and image sequences) and doing camera/motion tracking operations. - New editor called Clip Editor which is currently used for motion/tracking stuff only, but which can be easily extended to work with masks too. This editor supports: * Loading movie files/image sequences * Build proxies with different size for loaded movie clip, also supports building undistorted proxies to increase speed of playback in undistorted mode. * Manual lens distortion mode calibration using grid and grease pencil * Supervised 2D tracking using two different algorithms KLT and SAD. * Basic algorithm for feature detection * Camera motion solving. scene orientation - New constraints to "link" scene objects with solved motions from clip: * Follow Track (make object follow 2D motion of track with given name or parent object to reconstructed 3D position of track) * Camera Solver to make camera moving in the same way as reconstructed camera This commit NOT includes changes from tomato branch: - New nodes (they'll be commited as separated patch) - Automatic image offset guessing for image input node and image editor (need to do more tests and gather more feedback) - Code cleanup in libmv-capi. It's not so critical cleanup, just increasing readability and understanadability of code. Better to make this chaneg when Keir will finish his current patch. More details about this project can be found on this page: http://wiki.blender.org/index.php/User:Nazg-gul/GSoC-2011 Further development of small features would be done in trunk, bigger/experimental features would first be implemented in tomato branch.
2011-11-07 12:55:18 +00:00
/* try avoid unnecessary updates so pre-computed distortion grids are not freed */
Camera tracking integration =========================== Commiting camera tracking integration gsoc project into trunk. This commit includes: - Bundled version of libmv library (with some changes against official repo, re-sync with libmv repo a bit later) - New datatype ID called MovieClip which is optimized to work with movie clips (both of movie files and image sequences) and doing camera/motion tracking operations. - New editor called Clip Editor which is currently used for motion/tracking stuff only, but which can be easily extended to work with masks too. This editor supports: * Loading movie files/image sequences * Build proxies with different size for loaded movie clip, also supports building undistorted proxies to increase speed of playback in undistorted mode. * Manual lens distortion mode calibration using grid and grease pencil * Supervised 2D tracking using two different algorithms KLT and SAD. * Basic algorithm for feature detection * Camera motion solving. scene orientation - New constraints to "link" scene objects with solved motions from clip: * Follow Track (make object follow 2D motion of track with given name or parent object to reconstructed 3D position of track) * Camera Solver to make camera moving in the same way as reconstructed camera This commit NOT includes changes from tomato branch: - New nodes (they'll be commited as separated patch) - Automatic image offset guessing for image input node and image editor (need to do more tests and gather more feedback) - Code cleanup in libmv-capi. It's not so critical cleanup, just increasing readability and understanadability of code. Better to make this chaneg when Keir will finish his current patch. More details about this project can be found on this page: http://wiki.blender.org/index.php/User:Nazg-gul/GSoC-2011 Further development of small features would be done in trunk, bigger/experimental features would first be implemented in tomato branch.
2011-11-07 12:55:18 +00:00
if (camera_intrinsics->focal_length() != focal_length)
camera_intrinsics->SetFocalLength(focal_length, focal_length);
Camera tracking integration =========================== Commiting camera tracking integration gsoc project into trunk. This commit includes: - Bundled version of libmv library (with some changes against official repo, re-sync with libmv repo a bit later) - New datatype ID called MovieClip which is optimized to work with movie clips (both of movie files and image sequences) and doing camera/motion tracking operations. - New editor called Clip Editor which is currently used for motion/tracking stuff only, but which can be easily extended to work with masks too. This editor supports: * Loading movie files/image sequences * Build proxies with different size for loaded movie clip, also supports building undistorted proxies to increase speed of playback in undistorted mode. * Manual lens distortion mode calibration using grid and grease pencil * Supervised 2D tracking using two different algorithms KLT and SAD. * Basic algorithm for feature detection * Camera motion solving. scene orientation - New constraints to "link" scene objects with solved motions from clip: * Follow Track (make object follow 2D motion of track with given name or parent object to reconstructed 3D position of track) * Camera Solver to make camera moving in the same way as reconstructed camera This commit NOT includes changes from tomato branch: - New nodes (they'll be commited as separated patch) - Automatic image offset guessing for image input node and image editor (need to do more tests and gather more feedback) - Code cleanup in libmv-capi. It's not so critical cleanup, just increasing readability and understanadability of code. Better to make this chaneg when Keir will finish his current patch. More details about this project can be found on this page: http://wiki.blender.org/index.php/User:Nazg-gul/GSoC-2011 Further development of small features would be done in trunk, bigger/experimental features would first be implemented in tomato branch.
2011-11-07 12:55:18 +00:00
if (camera_intrinsics->principal_point_x() != principal_x ||
camera_intrinsics->principal_point_y() != principal_y)
{
camera_intrinsics->SetPrincipalPoint(principal_x, principal_y);
}
if (camera_intrinsics->k1() != k1 ||
camera_intrinsics->k2() != k2 ||
camera_intrinsics->k3() != k3)
{
camera_intrinsics->SetRadialDistortion(k1, k2, k3);
}
if (camera_intrinsics->image_width() != image_width ||
camera_intrinsics->image_height() != image_height)
{
camera_intrinsics->SetImageSize(image_width, image_height);
}
Camera tracking integration =========================== Commiting camera tracking integration gsoc project into trunk. This commit includes: - Bundled version of libmv library (with some changes against official repo, re-sync with libmv repo a bit later) - New datatype ID called MovieClip which is optimized to work with movie clips (both of movie files and image sequences) and doing camera/motion tracking operations. - New editor called Clip Editor which is currently used for motion/tracking stuff only, but which can be easily extended to work with masks too. This editor supports: * Loading movie files/image sequences * Build proxies with different size for loaded movie clip, also supports building undistorted proxies to increase speed of playback in undistorted mode. * Manual lens distortion mode calibration using grid and grease pencil * Supervised 2D tracking using two different algorithms KLT and SAD. * Basic algorithm for feature detection * Camera motion solving. scene orientation - New constraints to "link" scene objects with solved motions from clip: * Follow Track (make object follow 2D motion of track with given name or parent object to reconstructed 3D position of track) * Camera Solver to make camera moving in the same way as reconstructed camera This commit NOT includes changes from tomato branch: - New nodes (they'll be commited as separated patch) - Automatic image offset guessing for image input node and image editor (need to do more tests and gather more feedback) - Code cleanup in libmv-capi. It's not so critical cleanup, just increasing readability and understanadability of code. Better to make this chaneg when Keir will finish his current patch. More details about this project can be found on this page: http://wiki.blender.org/index.php/User:Nazg-gul/GSoC-2011 Further development of small features would be done in trunk, bigger/experimental features would first be implemented in tomato branch.
2011-11-07 12:55:18 +00:00
}
void libmv_CameraIntrinsicsSetThreads(struct libmv_CameraIntrinsics *libmv_intrinsics, int threads)
{
libmv::CameraIntrinsics *camera_intrinsics = (libmv::CameraIntrinsics *) libmv_intrinsics;
camera_intrinsics->SetThreads(threads);
}
void libmv_CameraIntrinsicsExtract(struct libmv_CameraIntrinsics *libmv_intrinsics, double *focal_length,
double *principal_x, double *principal_y, double *k1, double *k2, double *k3, int *width, int *height)
{
libmv::CameraIntrinsics *camera_intrinsics = (libmv::CameraIntrinsics *) libmv_intrinsics;
*focal_length = camera_intrinsics->focal_length();
*principal_x = camera_intrinsics->principal_point_x();
*principal_y = camera_intrinsics->principal_point_y();
*k1 = camera_intrinsics->k1();
*k2 = camera_intrinsics->k2();
Assorted camera tracker improvements - Add support for refining the camera's intrinsic parameters during a solve. Currently, refining supports only the following combinations of intrinsic parameters: f f, cx, cy f, cx, cy, k1, k2 f, k1 f, k1, k2 This is not the same as autocalibration, since the user must still make a reasonable initial guess about the focal length and other parameters, whereas true autocalibration would eliminate the need for the user specify intrinsic parameters at all. However, the solver works well with only rough guesses for the focal length, so perhaps full autocalibation is not that important. Adding support for the last two combinations, (f, k1) and (f, k1, k2) required changes to the library libmv depends on for bundle adjustment, SSBA. These changes should get ported upstream not just to libmv but to SSBA as well. - Improved the region of convergence for bundle adjustment by increasing the number of Levenberg-Marquardt iterations from 50 to 500. This way, the solver is able to crawl out of the bad local minima it gets stuck in when changing from, for example, bundling k1 and k2 to just k1 and resetting k2 to 0. - Add several new region tracker implementations. A region tracker is a libmv concept, which refers to tracking a template image pattern through frames. The impact to end users is that tracking should "just work better". I am reserving a more detailed writeup, and maybe a paper, for later. - Other libmv tweaks, such as detecting that a tracker is headed outside of the image bounds. This includes several changes made directly to the libmv extern code rather expecting to get those changes through normal libmv channels, because I, the libmv BDFL, decided it was faster to work on libmv directly in Blender, then later reverse-port the libmv changes from Blender back into libmv trunk. The interesting part is that I added a full Levenberg-Marquardt loop to the region tracking code, which should lead to a more stable solutions. I also added a hacky implementation of "Efficient Second-Order Minimization" for tracking, which works nicely. A more detailed quantitative evaluation will follow. Original patch by Keir, cleaned a bit by myself.
2011-11-14 06:41:23 +00:00
}
void libmv_CameraIntrinsicsUndistortByte(struct libmv_CameraIntrinsics *libmv_intrinsics,
Camera tracking integration =========================== Commiting camera tracking integration gsoc project into trunk. This commit includes: - Bundled version of libmv library (with some changes against official repo, re-sync with libmv repo a bit later) - New datatype ID called MovieClip which is optimized to work with movie clips (both of movie files and image sequences) and doing camera/motion tracking operations. - New editor called Clip Editor which is currently used for motion/tracking stuff only, but which can be easily extended to work with masks too. This editor supports: * Loading movie files/image sequences * Build proxies with different size for loaded movie clip, also supports building undistorted proxies to increase speed of playback in undistorted mode. * Manual lens distortion mode calibration using grid and grease pencil * Supervised 2D tracking using two different algorithms KLT and SAD. * Basic algorithm for feature detection * Camera motion solving. scene orientation - New constraints to "link" scene objects with solved motions from clip: * Follow Track (make object follow 2D motion of track with given name or parent object to reconstructed 3D position of track) * Camera Solver to make camera moving in the same way as reconstructed camera This commit NOT includes changes from tomato branch: - New nodes (they'll be commited as separated patch) - Automatic image offset guessing for image input node and image editor (need to do more tests and gather more feedback) - Code cleanup in libmv-capi. It's not so critical cleanup, just increasing readability and understanadability of code. Better to make this chaneg when Keir will finish his current patch. More details about this project can be found on this page: http://wiki.blender.org/index.php/User:Nazg-gul/GSoC-2011 Further development of small features would be done in trunk, bigger/experimental features would first be implemented in tomato branch.
2011-11-07 12:55:18 +00:00
unsigned char *src, unsigned char *dst, int width, int height, float overscan, int channels)
{
libmv::CameraIntrinsics *camera_intrinsics = (libmv::CameraIntrinsics *) libmv_intrinsics;
Camera tracking integration =========================== Commiting camera tracking integration gsoc project into trunk. This commit includes: - Bundled version of libmv library (with some changes against official repo, re-sync with libmv repo a bit later) - New datatype ID called MovieClip which is optimized to work with movie clips (both of movie files and image sequences) and doing camera/motion tracking operations. - New editor called Clip Editor which is currently used for motion/tracking stuff only, but which can be easily extended to work with masks too. This editor supports: * Loading movie files/image sequences * Build proxies with different size for loaded movie clip, also supports building undistorted proxies to increase speed of playback in undistorted mode. * Manual lens distortion mode calibration using grid and grease pencil * Supervised 2D tracking using two different algorithms KLT and SAD. * Basic algorithm for feature detection * Camera motion solving. scene orientation - New constraints to "link" scene objects with solved motions from clip: * Follow Track (make object follow 2D motion of track with given name or parent object to reconstructed 3D position of track) * Camera Solver to make camera moving in the same way as reconstructed camera This commit NOT includes changes from tomato branch: - New nodes (they'll be commited as separated patch) - Automatic image offset guessing for image input node and image editor (need to do more tests and gather more feedback) - Code cleanup in libmv-capi. It's not so critical cleanup, just increasing readability and understanadability of code. Better to make this chaneg when Keir will finish his current patch. More details about this project can be found on this page: http://wiki.blender.org/index.php/User:Nazg-gul/GSoC-2011 Further development of small features would be done in trunk, bigger/experimental features would first be implemented in tomato branch.
2011-11-07 12:55:18 +00:00
camera_intrinsics->Undistort(src, dst, width, height, overscan, channels);
Camera tracking integration =========================== Commiting camera tracking integration gsoc project into trunk. This commit includes: - Bundled version of libmv library (with some changes against official repo, re-sync with libmv repo a bit later) - New datatype ID called MovieClip which is optimized to work with movie clips (both of movie files and image sequences) and doing camera/motion tracking operations. - New editor called Clip Editor which is currently used for motion/tracking stuff only, but which can be easily extended to work with masks too. This editor supports: * Loading movie files/image sequences * Build proxies with different size for loaded movie clip, also supports building undistorted proxies to increase speed of playback in undistorted mode. * Manual lens distortion mode calibration using grid and grease pencil * Supervised 2D tracking using two different algorithms KLT and SAD. * Basic algorithm for feature detection * Camera motion solving. scene orientation - New constraints to "link" scene objects with solved motions from clip: * Follow Track (make object follow 2D motion of track with given name or parent object to reconstructed 3D position of track) * Camera Solver to make camera moving in the same way as reconstructed camera This commit NOT includes changes from tomato branch: - New nodes (they'll be commited as separated patch) - Automatic image offset guessing for image input node and image editor (need to do more tests and gather more feedback) - Code cleanup in libmv-capi. It's not so critical cleanup, just increasing readability and understanadability of code. Better to make this chaneg when Keir will finish his current patch. More details about this project can be found on this page: http://wiki.blender.org/index.php/User:Nazg-gul/GSoC-2011 Further development of small features would be done in trunk, bigger/experimental features would first be implemented in tomato branch.
2011-11-07 12:55:18 +00:00
}
void libmv_CameraIntrinsicsUndistortFloat(struct libmv_CameraIntrinsics *libmvIntrinsics,
float *src, float *dst, int width, int height, float overscan, int channels)
{
libmv::CameraIntrinsics *intrinsics = (libmv::CameraIntrinsics *) libmvIntrinsics;
intrinsics->Undistort(src, dst, width, height, overscan, channels);
}
void libmv_CameraIntrinsicsDistortByte(struct libmv_CameraIntrinsics *libmvIntrinsics,
unsigned char *src, unsigned char *dst, int width, int height, float overscan, int channels)
{
libmv::CameraIntrinsics *intrinsics = (libmv::CameraIntrinsics *) libmvIntrinsics;
intrinsics->Distort(src, dst, width, height, overscan, channels);
}
void libmv_CameraIntrinsicsDistortFloat(struct libmv_CameraIntrinsics *libmvIntrinsics,
float *src, float *dst, int width, int height, float overscan, int channels)
{
libmv::CameraIntrinsics *intrinsics = (libmv::CameraIntrinsics *) libmvIntrinsics;
intrinsics->Distort(src, dst, width, height, overscan, channels);
}
/* ************ utils ************ */
void libmv_applyCameraIntrinsics(libmv_cameraIntrinsicsOptions *libmv_camera_intrinsics_options,
double x, double y, double *x1, double *y1)
Camera tracking integration =========================== Commiting camera tracking integration gsoc project into trunk. This commit includes: - Bundled version of libmv library (with some changes against official repo, re-sync with libmv repo a bit later) - New datatype ID called MovieClip which is optimized to work with movie clips (both of movie files and image sequences) and doing camera/motion tracking operations. - New editor called Clip Editor which is currently used for motion/tracking stuff only, but which can be easily extended to work with masks too. This editor supports: * Loading movie files/image sequences * Build proxies with different size for loaded movie clip, also supports building undistorted proxies to increase speed of playback in undistorted mode. * Manual lens distortion mode calibration using grid and grease pencil * Supervised 2D tracking using two different algorithms KLT and SAD. * Basic algorithm for feature detection * Camera motion solving. scene orientation - New constraints to "link" scene objects with solved motions from clip: * Follow Track (make object follow 2D motion of track with given name or parent object to reconstructed 3D position of track) * Camera Solver to make camera moving in the same way as reconstructed camera This commit NOT includes changes from tomato branch: - New nodes (they'll be commited as separated patch) - Automatic image offset guessing for image input node and image editor (need to do more tests and gather more feedback) - Code cleanup in libmv-capi. It's not so critical cleanup, just increasing readability and understanadability of code. Better to make this chaneg when Keir will finish his current patch. More details about this project can be found on this page: http://wiki.blender.org/index.php/User:Nazg-gul/GSoC-2011 Further development of small features would be done in trunk, bigger/experimental features would first be implemented in tomato branch.
2011-11-07 12:55:18 +00:00
{
libmv::CameraIntrinsics camera_intrinsics;
Camera tracking integration =========================== Commiting camera tracking integration gsoc project into trunk. This commit includes: - Bundled version of libmv library (with some changes against official repo, re-sync with libmv repo a bit later) - New datatype ID called MovieClip which is optimized to work with movie clips (both of movie files and image sequences) and doing camera/motion tracking operations. - New editor called Clip Editor which is currently used for motion/tracking stuff only, but which can be easily extended to work with masks too. This editor supports: * Loading movie files/image sequences * Build proxies with different size for loaded movie clip, also supports building undistorted proxies to increase speed of playback in undistorted mode. * Manual lens distortion mode calibration using grid and grease pencil * Supervised 2D tracking using two different algorithms KLT and SAD. * Basic algorithm for feature detection * Camera motion solving. scene orientation - New constraints to "link" scene objects with solved motions from clip: * Follow Track (make object follow 2D motion of track with given name or parent object to reconstructed 3D position of track) * Camera Solver to make camera moving in the same way as reconstructed camera This commit NOT includes changes from tomato branch: - New nodes (they'll be commited as separated patch) - Automatic image offset guessing for image input node and image editor (need to do more tests and gather more feedback) - Code cleanup in libmv-capi. It's not so critical cleanup, just increasing readability and understanadability of code. Better to make this chaneg when Keir will finish his current patch. More details about this project can be found on this page: http://wiki.blender.org/index.php/User:Nazg-gul/GSoC-2011 Further development of small features would be done in trunk, bigger/experimental features would first be implemented in tomato branch.
2011-11-07 12:55:18 +00:00
cameraIntrinsicsFromOptions(&camera_intrinsics, libmv_camera_intrinsics_options);
Camera tracking integration =========================== Commiting camera tracking integration gsoc project into trunk. This commit includes: - Bundled version of libmv library (with some changes against official repo, re-sync with libmv repo a bit later) - New datatype ID called MovieClip which is optimized to work with movie clips (both of movie files and image sequences) and doing camera/motion tracking operations. - New editor called Clip Editor which is currently used for motion/tracking stuff only, but which can be easily extended to work with masks too. This editor supports: * Loading movie files/image sequences * Build proxies with different size for loaded movie clip, also supports building undistorted proxies to increase speed of playback in undistorted mode. * Manual lens distortion mode calibration using grid and grease pencil * Supervised 2D tracking using two different algorithms KLT and SAD. * Basic algorithm for feature detection * Camera motion solving. scene orientation - New constraints to "link" scene objects with solved motions from clip: * Follow Track (make object follow 2D motion of track with given name or parent object to reconstructed 3D position of track) * Camera Solver to make camera moving in the same way as reconstructed camera This commit NOT includes changes from tomato branch: - New nodes (they'll be commited as separated patch) - Automatic image offset guessing for image input node and image editor (need to do more tests and gather more feedback) - Code cleanup in libmv-capi. It's not so critical cleanup, just increasing readability and understanadability of code. Better to make this chaneg when Keir will finish his current patch. More details about this project can be found on this page: http://wiki.blender.org/index.php/User:Nazg-gul/GSoC-2011 Further development of small features would be done in trunk, bigger/experimental features would first be implemented in tomato branch.
2011-11-07 12:55:18 +00:00
if (libmv_camera_intrinsics_options->focal_length) {
Camera tracking integration =========================== Commiting camera tracking integration gsoc project into trunk. This commit includes: - Bundled version of libmv library (with some changes against official repo, re-sync with libmv repo a bit later) - New datatype ID called MovieClip which is optimized to work with movie clips (both of movie files and image sequences) and doing camera/motion tracking operations. - New editor called Clip Editor which is currently used for motion/tracking stuff only, but which can be easily extended to work with masks too. This editor supports: * Loading movie files/image sequences * Build proxies with different size for loaded movie clip, also supports building undistorted proxies to increase speed of playback in undistorted mode. * Manual lens distortion mode calibration using grid and grease pencil * Supervised 2D tracking using two different algorithms KLT and SAD. * Basic algorithm for feature detection * Camera motion solving. scene orientation - New constraints to "link" scene objects with solved motions from clip: * Follow Track (make object follow 2D motion of track with given name or parent object to reconstructed 3D position of track) * Camera Solver to make camera moving in the same way as reconstructed camera This commit NOT includes changes from tomato branch: - New nodes (they'll be commited as separated patch) - Automatic image offset guessing for image input node and image editor (need to do more tests and gather more feedback) - Code cleanup in libmv-capi. It's not so critical cleanup, just increasing readability and understanadability of code. Better to make this chaneg when Keir will finish his current patch. More details about this project can be found on this page: http://wiki.blender.org/index.php/User:Nazg-gul/GSoC-2011 Further development of small features would be done in trunk, bigger/experimental features would first be implemented in tomato branch.
2011-11-07 12:55:18 +00:00
/* do a lens undistortion if focal length is non-zero only */
camera_intrinsics.ApplyIntrinsics(x, y, x1, y1);
Camera tracking integration =========================== Commiting camera tracking integration gsoc project into trunk. This commit includes: - Bundled version of libmv library (with some changes against official repo, re-sync with libmv repo a bit later) - New datatype ID called MovieClip which is optimized to work with movie clips (both of movie files and image sequences) and doing camera/motion tracking operations. - New editor called Clip Editor which is currently used for motion/tracking stuff only, but which can be easily extended to work with masks too. This editor supports: * Loading movie files/image sequences * Build proxies with different size for loaded movie clip, also supports building undistorted proxies to increase speed of playback in undistorted mode. * Manual lens distortion mode calibration using grid and grease pencil * Supervised 2D tracking using two different algorithms KLT and SAD. * Basic algorithm for feature detection * Camera motion solving. scene orientation - New constraints to "link" scene objects with solved motions from clip: * Follow Track (make object follow 2D motion of track with given name or parent object to reconstructed 3D position of track) * Camera Solver to make camera moving in the same way as reconstructed camera This commit NOT includes changes from tomato branch: - New nodes (they'll be commited as separated patch) - Automatic image offset guessing for image input node and image editor (need to do more tests and gather more feedback) - Code cleanup in libmv-capi. It's not so critical cleanup, just increasing readability and understanadability of code. Better to make this chaneg when Keir will finish his current patch. More details about this project can be found on this page: http://wiki.blender.org/index.php/User:Nazg-gul/GSoC-2011 Further development of small features would be done in trunk, bigger/experimental features would first be implemented in tomato branch.
2011-11-07 12:55:18 +00:00
}
}
void libmv_InvertIntrinsics(libmv_cameraIntrinsicsOptions *libmv_camera_intrinsics_options,
double x, double y, double *x1, double *y1)
Camera tracking integration =========================== Commiting camera tracking integration gsoc project into trunk. This commit includes: - Bundled version of libmv library (with some changes against official repo, re-sync with libmv repo a bit later) - New datatype ID called MovieClip which is optimized to work with movie clips (both of movie files and image sequences) and doing camera/motion tracking operations. - New editor called Clip Editor which is currently used for motion/tracking stuff only, but which can be easily extended to work with masks too. This editor supports: * Loading movie files/image sequences * Build proxies with different size for loaded movie clip, also supports building undistorted proxies to increase speed of playback in undistorted mode. * Manual lens distortion mode calibration using grid and grease pencil * Supervised 2D tracking using two different algorithms KLT and SAD. * Basic algorithm for feature detection * Camera motion solving. scene orientation - New constraints to "link" scene objects with solved motions from clip: * Follow Track (make object follow 2D motion of track with given name or parent object to reconstructed 3D position of track) * Camera Solver to make camera moving in the same way as reconstructed camera This commit NOT includes changes from tomato branch: - New nodes (they'll be commited as separated patch) - Automatic image offset guessing for image input node and image editor (need to do more tests and gather more feedback) - Code cleanup in libmv-capi. It's not so critical cleanup, just increasing readability and understanadability of code. Better to make this chaneg when Keir will finish his current patch. More details about this project can be found on this page: http://wiki.blender.org/index.php/User:Nazg-gul/GSoC-2011 Further development of small features would be done in trunk, bigger/experimental features would first be implemented in tomato branch.
2011-11-07 12:55:18 +00:00
{
libmv::CameraIntrinsics camera_intrinsics;
Camera tracking integration =========================== Commiting camera tracking integration gsoc project into trunk. This commit includes: - Bundled version of libmv library (with some changes against official repo, re-sync with libmv repo a bit later) - New datatype ID called MovieClip which is optimized to work with movie clips (both of movie files and image sequences) and doing camera/motion tracking operations. - New editor called Clip Editor which is currently used for motion/tracking stuff only, but which can be easily extended to work with masks too. This editor supports: * Loading movie files/image sequences * Build proxies with different size for loaded movie clip, also supports building undistorted proxies to increase speed of playback in undistorted mode. * Manual lens distortion mode calibration using grid and grease pencil * Supervised 2D tracking using two different algorithms KLT and SAD. * Basic algorithm for feature detection * Camera motion solving. scene orientation - New constraints to "link" scene objects with solved motions from clip: * Follow Track (make object follow 2D motion of track with given name or parent object to reconstructed 3D position of track) * Camera Solver to make camera moving in the same way as reconstructed camera This commit NOT includes changes from tomato branch: - New nodes (they'll be commited as separated patch) - Automatic image offset guessing for image input node and image editor (need to do more tests and gather more feedback) - Code cleanup in libmv-capi. It's not so critical cleanup, just increasing readability and understanadability of code. Better to make this chaneg when Keir will finish his current patch. More details about this project can be found on this page: http://wiki.blender.org/index.php/User:Nazg-gul/GSoC-2011 Further development of small features would be done in trunk, bigger/experimental features would first be implemented in tomato branch.
2011-11-07 12:55:18 +00:00
cameraIntrinsicsFromOptions(&camera_intrinsics, libmv_camera_intrinsics_options);
Camera tracking integration =========================== Commiting camera tracking integration gsoc project into trunk. This commit includes: - Bundled version of libmv library (with some changes against official repo, re-sync with libmv repo a bit later) - New datatype ID called MovieClip which is optimized to work with movie clips (both of movie files and image sequences) and doing camera/motion tracking operations. - New editor called Clip Editor which is currently used for motion/tracking stuff only, but which can be easily extended to work with masks too. This editor supports: * Loading movie files/image sequences * Build proxies with different size for loaded movie clip, also supports building undistorted proxies to increase speed of playback in undistorted mode. * Manual lens distortion mode calibration using grid and grease pencil * Supervised 2D tracking using two different algorithms KLT and SAD. * Basic algorithm for feature detection * Camera motion solving. scene orientation - New constraints to "link" scene objects with solved motions from clip: * Follow Track (make object follow 2D motion of track with given name or parent object to reconstructed 3D position of track) * Camera Solver to make camera moving in the same way as reconstructed camera This commit NOT includes changes from tomato branch: - New nodes (they'll be commited as separated patch) - Automatic image offset guessing for image input node and image editor (need to do more tests and gather more feedback) - Code cleanup in libmv-capi. It's not so critical cleanup, just increasing readability and understanadability of code. Better to make this chaneg when Keir will finish his current patch. More details about this project can be found on this page: http://wiki.blender.org/index.php/User:Nazg-gul/GSoC-2011 Further development of small features would be done in trunk, bigger/experimental features would first be implemented in tomato branch.
2011-11-07 12:55:18 +00:00
if (libmv_camera_intrinsics_options->focal_length) {
Camera tracking integration =========================== Commiting camera tracking integration gsoc project into trunk. This commit includes: - Bundled version of libmv library (with some changes against official repo, re-sync with libmv repo a bit later) - New datatype ID called MovieClip which is optimized to work with movie clips (both of movie files and image sequences) and doing camera/motion tracking operations. - New editor called Clip Editor which is currently used for motion/tracking stuff only, but which can be easily extended to work with masks too. This editor supports: * Loading movie files/image sequences * Build proxies with different size for loaded movie clip, also supports building undistorted proxies to increase speed of playback in undistorted mode. * Manual lens distortion mode calibration using grid and grease pencil * Supervised 2D tracking using two different algorithms KLT and SAD. * Basic algorithm for feature detection * Camera motion solving. scene orientation - New constraints to "link" scene objects with solved motions from clip: * Follow Track (make object follow 2D motion of track with given name or parent object to reconstructed 3D position of track) * Camera Solver to make camera moving in the same way as reconstructed camera This commit NOT includes changes from tomato branch: - New nodes (they'll be commited as separated patch) - Automatic image offset guessing for image input node and image editor (need to do more tests and gather more feedback) - Code cleanup in libmv-capi. It's not so critical cleanup, just increasing readability and understanadability of code. Better to make this chaneg when Keir will finish his current patch. More details about this project can be found on this page: http://wiki.blender.org/index.php/User:Nazg-gul/GSoC-2011 Further development of small features would be done in trunk, bigger/experimental features would first be implemented in tomato branch.
2011-11-07 12:55:18 +00:00
/* do a lens distortion if focal length is non-zero only */
camera_intrinsics.InvertIntrinsics(x, y, x1, y1);
Camera tracking integration =========================== Commiting camera tracking integration gsoc project into trunk. This commit includes: - Bundled version of libmv library (with some changes against official repo, re-sync with libmv repo a bit later) - New datatype ID called MovieClip which is optimized to work with movie clips (both of movie files and image sequences) and doing camera/motion tracking operations. - New editor called Clip Editor which is currently used for motion/tracking stuff only, but which can be easily extended to work with masks too. This editor supports: * Loading movie files/image sequences * Build proxies with different size for loaded movie clip, also supports building undistorted proxies to increase speed of playback in undistorted mode. * Manual lens distortion mode calibration using grid and grease pencil * Supervised 2D tracking using two different algorithms KLT and SAD. * Basic algorithm for feature detection * Camera motion solving. scene orientation - New constraints to "link" scene objects with solved motions from clip: * Follow Track (make object follow 2D motion of track with given name or parent object to reconstructed 3D position of track) * Camera Solver to make camera moving in the same way as reconstructed camera This commit NOT includes changes from tomato branch: - New nodes (they'll be commited as separated patch) - Automatic image offset guessing for image input node and image editor (need to do more tests and gather more feedback) - Code cleanup in libmv-capi. It's not so critical cleanup, just increasing readability and understanadability of code. Better to make this chaneg when Keir will finish his current patch. More details about this project can be found on this page: http://wiki.blender.org/index.php/User:Nazg-gul/GSoC-2011 Further development of small features would be done in trunk, bigger/experimental features would first be implemented in tomato branch.
2011-11-07 12:55:18 +00:00
}
}