- Z-axis rotation limits were not working, was using wrong flag.
- Don't allow min limit to go over max, or vice versa.
- Fix for jacobian getting overwritten with IK clamping.
Fix some warnings.
Merge with latest soc code.
What changed in IK lib:
Fully restructured, with components now as follows:
- IK_Solver: C <=> C++ interface
- IK_QSegment: base class for bone/segment with 0
to 3 DOF
- IK_QTask: base class for a task (currently there's
a position and a rotation task)
- IK_QJacobian: the Jacobian matrix, with SVD
decomposition, damping, etc
- IK_QJacobianSolver: the iterative solver
The exponential map parametrization is no longer used,
instead we have now:
- 3DOF and 2DOF XZ segments: directly update matrix
with Rodrigues' formula
- Other: Euler angles (no worries about singularities
here)
Computation of the Jacobian inverse has also changed:
- The SVD algorithm is now based on LAPACK code,
instead of NR, to avoid some problems with rounding
errors.
- When the problem is underconstrained (as is the case
most of the time), the SVD is computed for the transpose
of the Jacobian (faster).
- A new damping algorithm called the Selectively Damped
Least Squares is used, result in faster and more
stable convergence.
- Stiffness is implemented as if a weighted psuedo-inverse
was used.
Tree structure support.
Rotation limits:
- 3DOF and 2DOF XZ segments limits are based on a swing
(direct axis-angle over XZ) and twist/roll (rotation
over Y) decomposition. The swing region is an ellipse
on a sphere.
- Rotation limits are implemented using an inner clamping
loop: as long as there is a violation, a violating DOF
is clamped and removed from the Jacobian, and the solution
is recomputed.
Convergence checking is based now on the max norm of angle
change, or the maximum number of iterations.
64 bits systems... weird bug. :)
It now only does a unit-of-4 check, for all systems. This will work fine,
since the malloc code will return aligned anyway, and the guarded alloc
system only stores ints in the headers. Also, the sizeof() call will
correctly do padding, so there's no risk of allocating too small blocks.
- XCode project
- Scons :
scons stopped working for Os X in the last month (dont know when) :
* the '.' and '..' keywords in CCPPATH
are not recognized anymore for sconscripts compiling files at
a sub level
* when doing a scons clean, the 3 subdirs in build dir are removed
and scons then fail to recreate them (exten, intern, source)
this commit solve the first problem as a temp workaround
for the latter simply recreate the dirs manually
I will investigate that further when on vacations, which i should already
be.
;(
all 3 build systems are now working on Os X
errors, switched MEM_set_error_stream to MEM_set_error_callback
that calls a function to print result instead of just giving
a FILE *
Note: requires intern recompile
Code provided by Matthew Plough. Verified to run OK in OSX, Windows.
Solves bug #2598, which crashed blender on using non-manifold meshes with
decimator.
source/creator/creator.c
changed ifdef's around fpe_handler to match when its actually used.
intern/SoundSystem/intern/SND_AudioDevice.cpp
Changed: SND_SoundObject* oldobject = oldobject = pIdObject->GetSoundObject();
to: SND_SoundObject* oldobject = pIdObject->GetSoundObject();
intern/SoundSystem/openal/SND_OpenALDevice.cpp
removed unused var.
source/blender/blenkernel/intern/mball.c
initalized a couple of vars that might have been used uninitalized.
The rest were changing types to match, most of them were something like
was short * should have been unsigned short *.
Kent
Added the is_a_really_crappy_nvidia_card() call in BMF_DrawString(), this
to solve a bug in NVidia 6800 drivers of MacOSX G5. It is #ifdeffed for
OSX only, and queries for a NVidia 6800 card to activate the patch.
The issue is that these drivers forgot to correctly implement viewport()
offset for drawing bitmap fonts, causing text display in Blender to be
invisible, except for the leftmost/bottom sub window.
This hack will be removed when Apple releases a driver upgrade, which is
unknown when to happen. Has to be decided still if this is worth for a
release, or that we provide the hack as separate download.
Thanks Randall Rickert for all testing, and Daniel for code review! :)
I got rid of a few warnings about blah shadows a previous declaration.
In the gameengine files I fix the following:
removed some unused vars
removed dos style line breaks
added newlines to last line in a couple of files to remove warnings.
Kent
- Added correct .libs to be linked to GP_Ghost
- Fixed incorrect library extensions for the linker (*.lib and not unix *.a)
- Projects have been set to be multi-threaded - projects were being linked to different versions of the CRT
- All projects are now linking to python 2.3 (this fixes an error regarding struct sizes in python)
- Removed invalid source paths from inclusion for the compiler
- Fix dependency issues with the GP_Ghost module
- Proper .dll's are now being copied to their respective locations for debug and release version of blender/blenderplayer
- Warnings have been set to level 2
- Import paths have been updated and organized for the compiler
- Duplicate import paths for the compiler have been removed
- Removed silly silly silly silly unix path-separators from window paths
- Post-Build scripts have been updated where neccessary (XCOPY is retatined)
- Fixed the output of various .lib files (PHY_Sumo and Yafray)
- PHY_Sumo now uses a .pch
- Removed #undef _DEBUG from debug builds :p
- FTF_Font is now contains a debug build and is being set to it's correct output path
- Blender/Blenderplayer/3DPlugin all linkto Debug libs where appropriate (instead of static release libs to avoid conflicts with the CRT)
- Python23_d.dll is now loaded into a true debug output of both blenderplayer and blender debug builds
- msvcrtd.dll is now loaded into a true debug output of both blenderplayer and blender debug builds
- Post-builds now clean up their mess
- ActiveX controller now builds
- BlenderPlayer now builds
- Internal libs now using .pch
- Moved the build directory from /blender/obj to ../build/msvc_7/ to follow precedence with scons builds
- Builds are now organized to follow precedence with scons builds with a /libs and a /intern /extern /source obj folders
- Set the proper build path for opennl
- Forcing BLO_loader source files to follow project defaults
- Added an 'extern' project that mimics the /make build of the 'intern' project
- Extern build includes solid and it's associated builds, ftgl, and freetype
- Added .vcproj files to /extern directory
- Added new extern.sln file to /extern directory
- All project now link to the built solid.lib from the extern project instead of lib/windows
- Removed uneeded XCOPYs commands from post-builds
- Added conditional Directory building to post-builds
- Freetype2 source commit (this is need for the building of a correct MFC linked version of the activeX controller as well as a correct release and debug build of the blender project)
- Ftgl is built and linked by the 'extern' project
- Intern/Extern project are now exporting their proper header files
- PHY_Physics is now linking to the correct set of header files (in build path not source)
- Makesdna builds object files into build directory now
- Projects are now ignoring the CORRECT versions of the CRT's :)
- Renderconverter is now outputting it's obj file to it's own directory instead of /render
- Gen_system is now being built to ../kernel
- BL_src_cre is now being built to /creator
- *.exp and *.ilk are now being cleaned up from linkers
- Libpng.dll and zlib.dll are now being copied to the /bin
- MTDLL build are now available for the correct projects
- ActiveX 3D Plugin now links correctly to solid and freetype2
- Corrected the Preprocessor #define that was labeled incorrectly in FTF_Font and Blender that was causing the FTGL libs to export dll calling conventions.
- Changed the solution configurations to 3D Plugin Release, Debug and Blender Release, Debug for internal and external librarys
- Changed the solution configurations to 3D Plugin Release, Debug and Blender Release, Debug for blender and added 2 new solutions configurations BlenderPlayer Release, Debug
-
- set build_install_all as the default project (if you give me a quarter ill tell you how :)
- set SoundSystem to it's proper dependencies
- updated those silly silly unix path separators
- set project warning levels to something not so noisy
It will attempt to break into the debugger instead of aborting the program. On Windows you have a nice MessageBox function, so you can choose to break, ignore, or permanently ignore the assert.
Floating point imprecision made MT_Quaternion::angle return NaN, since acos(x) is NaN for |x| > 1.
Because of the way NaN's propagate through float math, the view pos would be set to [NaN, NaN, NaN] resulting in a grey screen.
behaves very unpredictable (ask google) and seems to clear pending events
as well (fly mode, shift event dissappears).
The now used call, CGWarpMouseCursorPosition, creates no event so it has
to be combined with updating mouse cursor location in ghostwinlay.c
under linux / x11 / KDE.
The new behaviour in 2.34, not recreating the render window on re-render,
revealed that raising windows did not work under some window managers. Now the
"net wm extensions" are used if available, otherwise it will work the same way
as before (e.g. Gtk+ does this as well).
More info:
http://www.freedesktop.org/Standards/wm-spec
Caught from the following warning:
In file included from IK_QChain.h:47,
from IK_QChain.cpp:44:
TNT/cmat.h: In function `void TNT::matmultdiag(TNT::Matrix<T>&, const TNT::Matri
x<T>&, const TNT::Vector<T>&)':
TNT/cmat.h:593: warning: left-hand operand of comma has no effect
Kent
those who work with visible dock where experimenting slowdown of UI when
blender window was overlapping dock.
now :
- check available space excluding dock
- create window with a 10 pixels border
- maximised mode is only 1 click away
This code is fixing only. Will review that when consensus will be found
about how we should create the window.
if video card is open GL accelerated and has 16 Mo or more
start window in maximized mode wich is a full screen mode
but keeping access to other windows and sytem menu items
older comps start as usual
intern/SoundSystem/intern/SND_AudioDevice.cpp
Initalized a var that could fall through with no value.
source/blender/readblenfile/intern/BLO_readblenfile.c
source/blender/src/space.c
intern/SoundSystem/intern/SND_Scene.cpp
source/gameengine/Converter/BL_MeshDeformer.cpp
removed unused var's
intern/SoundSystem/openal/pthread_cancel.cpp
fixed a nested /*
source/blender/imbuf/IMB_imbuf.h
added static to the type returned for addzbufImBuf
source/blender/imbuf/intern/IMB_bmp.h
had a wrong prototype
source/blender/src/view.c
added newline at end of file.
source/blender/src/sequence.c
removed unused var
added #include <stdlib.h> to avoid:
warning: implicit declaration of function `abs'
initalized a var that could have been used without being set.
Kent
http://crd.lbl.gov/~xiaoye/SuperLU/
This is a library to solve sparse matrix systems (type A*x=B). It is able
to solve large systems very FAST. Only the necessary parts of the library
are included to limit file size and compilation time. This means the example
files, fortran interface, test files, matlab interface, cblas library,
complex number part and build system have been left out. All (gcc) warnings
have been fixed too.
This library will be used for LSCM UV unwrapping. With this library, LSCM
unwrapping can be calculated in a split second, making the unwrapping proces
much more interactive.
Added OpenNL (Open Numerical Libary):
http://www.loria.fr/~levy/OpenNL/
OpenNL is a library to easily construct and solve sparse linear systems. We
use a stripped down version, as an interface to SuperLU.
This library was kindly given to use by Bruno Levy.
So now by default, assert is not used in moto code.
If you want to use assert (and stop blender if one fail),
define MT_DEBUG.
solve bug #1292 and better fix for #1260
as there is a conflict between openal and sdl (when compiled with cdrom
support), i had to disable support for cd playing using sdl on os x
for the time being.
Now we're on the root of the problem. Ghost (OSX) didn't accept rawkeys
for the 1-2-...0 keys on a french keyboard. These have apparently different
symbols on it. (these silly french! :)
This commit intercepts the rawkeys in convertKey() call, and manually
makes sure theyre correctly mapped.
So: now french (spanish, etc) OSX users can finally use layer hotkeys! And
the popup menu shortcuts! :)
Tested & verified by Lukep. Merci!
var instead) return (float)atof(something);
is now float x=atof(something); return x;
There is still another one(warning)in this file but hey its a start.
Kent
* Blender static now links. By default this option is disabled on all
platforms. Simply set the option in config.opts to 'true'.
* Added the following flags to config.opts:
- HOST_CC. This is the C compiler for the host platform. This value is the
same as TARGET_CC when not cross compiling.
- HOST_CXX. This is the C++ compiler for the host platform. This value is
the same as TARGET_CXX when not cross compiling.
- TARGET_CC. This is the C compiler for the target platform.
- TARGET_CXX. This is the C++ compiler for the target platform.
- TARGET_AR. This is the linker command for linking libraries.
- PATH This is the standard search path
All SConscript files have been updated to reflect these changes. Now it's
possible to change only the root SConstruct file, and all compiler specific
variables are passed automatically to all SConscript files. Of course, this
does not apply to makesdna because there the host and target platform is
different from all other libraries.
To pass a variable that applies to all platforms, all we now have to do is
set the correct value in library_env
Note: as usual, to get the latest options in the config.opts file, first
remove your version.
* OpenAL support is now available on Linux. By default this feature is
disabled since it is only necessary when building the game engine. And the
game engine is disabled by default as well, so...
* Added 3 configurable options to config.opts. These apply to OpenAL settings.
NOTE: remove your current config.opts file to get the new options. (remember
your current settings though ;) )
2nd NOTE: All options* are now configurable via the config.opts file. If the
default settings for your platform are not correct, you should be
able to only update the config.opts file.
* FMOD is still not available - and I do not know if it is necessary to
enable this feature.
* libraries are now generated in [BUILD_DIR]/lib
* passed the user_options to all libraries now.
This means I could remove a couple of Export/Import lines.
* Changed the order in source/blender/src/SConscript and
source/gameengine/SConscript.
All libraries are now sorted alphabetically. This has no impact on the build
process.
- Fixed up the paths for "/lib/..." in csg.dsp
Quick note: Currently intern still builds the old lib (bsplib).
To compile the new lib (when everything is ready for it), uncheck
bsplib from the dependancies in build_install_all and replace with csg.
Will eventually drop bsplib completely - but not just yet :)
