of closures limit. Optimized the code now so it can handle more.
Change SVM mix/add closure handling, now we transform the node graph so that
the mix weights are fed into the closure nodes directly.
* Rename Legacy to Standard, it's not being deprecated as far as I know.
* Make option to toggle off Location solving work with Standard.
* Make it converge a bit better in some cases by enforcing a minimum number of
iterations before giving up.
* Move IK solver choice out of bone panel, it's an armature level setting and
should be set there.
* MEM_CacheLimitier - Size type to int conversion, should be safe for now (doing my best Bill Gates 640k impression)
* OpenNL CMakeLists.txt - MSVC and GCC have slightly different ways to remove definitions (DEBUG) without the compiler complaining
* BLI_math inlines - The include guard name and inline option macro name should be different. Suppressed warning about not exporting any symbols from inline math library
* BLI string / utf8 - Fixed some inconsistencies between declarations and definitions
* nodes - node_composite_util is apparently not used unless you enable the legacy compositor, so it should not be compiled in that case.
Leaving out changes to BLI_fileops for now, need to do more testing.
Also fix:
- Fluid simulation was always lagging 1 frame behind: E.g. the 250th frame in blender showed 249th frame of the fluid simulation.
Change:
- Animated enabled/disabled property only gets counted as "on" if value >= 1
Note that this bugfix should solve many problems with timings of animated fluid sim properties.
Now one no longer needs to match the sensor dimensions with the render dimensions manually.
IMPORTANT NOTE: if you were using AUTO before with mismathing sensor aspect ratio (comparing to the render dimensions)
this will change your render! We can doversion this, but apart from Tube project I don't know if anyone else
is using this yet (part due to this bug and the only recently fixed 3dview preview aspect ratio).
That should help more artists to take advantage of this fantastic Blender feature.
It still helps to know the parameters of kwnown cameras/lens though.
For example:
Nikon DX2S with a 10.5mm fisheye can be set with:
Render resolution: 4288 x 2848
Sensor 23.7 x 15.70 (15.70 can be ommitted if AUTO is used as fit method)
Note: some cameras render different sizes according to the recording mode.
For example, a Red Scarlet in 5k (@12 fps) can render a full circular fisheye with a sigma 4.5 lens.
The same camera in the 30fps recording mode renders 4k in a cropped circular image.
So it's not only the resolution that changes, but the actual sensor been used.
So just keep in mind that the more information you have from the camera/lens you want to emulate the better.
Bug found at/patch written as a follow up of the BlenderPRO2012, patch reviewed by Brecht Van Lommel
Linking happens fine, but blender crashes on startup -- crash with familiar
backtrace happens with i18n disabled (in that case it's something to do with OIIO).
This commit makes BKE_image_acquire_ibuf referencing result, which means once
some area requested for image buffer, it'll be guaranteed this buffer wouldn't
be freed by image signal.
To de-reference buffer BKE_image_release_ibuf should now always be used.
To make referencing working correct we can not rely on result of
image_get_ibuf_threadsafe called outside from thread lock. This is so because
we need to guarantee getting image buffer from list of loaded buffers and it's
referencing happens atomic. Without lock here it is possible that between call
of image_get_ibuf_threadsafe and referencing the buffer IMA_SIGNAL_FREE would
be called. Image signal handling too is blocking now to prevent such a
situation.
Threads are locking by spinlock, which are faster than mutexes. There were some
slowdown reports in the past about render slowdown when using OSX on Xeon CPU.
It shouldn't happen with spin locks, but more tests on different hardware would
be really welcome. So far can not see speed regressions on own computers.
This commit also removes BKE_image_get_ibuf, because it was not so intuitive
when get_ibuf and acquire_ibuf should be used.
Thanks to Ton and Brecht for discussion/review :)
Initial support of OSL builds using SCons build system. Only tested on Linux now.
No changes to configuration files themselves -- for now check how it's configured
for linux buildbot (it was already horror to make all this changes and verify them,
changes to linux-config.py could easily be done later).
Currently WITH_BF_STATICOSL and WITH_BF_STATICLLVM are more like rudiments because
linking against oslexec requires special trick with --whole-archive. We woul either
need to find a way dealing with this oslexec less hackish or drop STATICOSL and
STATICLLVM flags. Will keep dropping this flags for until we have "final" build
rules for OSL.
Still can not make 32bit linux rendering with OSL -- blender simply crashes when
starting rendering. So for time being this issues are solving disabled OSL for
32bit build slaves.
This commit adds a small and simplistic C wrapper around boost's locale library as intern/locale, and heavily simplifies/reduces Blender's own i18n code (under blenfont/ dir). And it adds back UI translation on windows' official builds (with msvc)!
Note to platform maintainers: iconv and gettext (libintl) can now be removed from precompiled libs (not gettext binaries, under windows, of course ;) ).
Note to MinGW32/64 users: boost_locale lib has not yet been uploaded for those build env, please disable WITH_INTERNATIONAL for now (hopefully will be fixed very soon, have contacted psy-fy).
problem is that the curved interpolation is not constant speed which leads to
mismatches. Turns out this is really hard to solve and implement efficiently, so
curved motion blur is disabled for now, until I can find a solution.
compute_device_list is using static vector of device information which
had pointers (identifier and name) to values from device information
structures. That structures are also stored in static vector and being
refreshed every 5 seconds.
The issue is, as soon as device information is being updated, pointers
in vector from compute_device_list became incorrect.
Seems it was the reason of issues with sudden switching from CUDA to
OpenCL on my desktop and from CUDA to CPU on my laptop, It was also
seems to be making persistent images behaves instable.
Made it so device identifier and name are copied from device info
to structures used by RNA (CCLDeviceInfo).
Alternative could be avoid cacheing CCLDeviceInfo and always use actual
list of device information by RNA. It shouldn't be so much slow.
was added for cycles.
This fixes the case where the option is disabled. I moved the option now to
Blender itself and made it keep the engine around only when it's enabled. Also
fixes case where there could be issues when switching to another renderer.
r = lens * theta
Thanks for Adriano Oliveira for reporting this and chasing down the right formula.
Now fulldome works no longer need to use equisolid + a specific lens+sensor size.
And happy birthday to me. And yes, that's how I celebrate it ;)
This option enables keeping loaded images in the memory in-between
of rendering.
Implemented by keeping render engine alive for until Render structure
is being freed.
Cycles will free all data when render finishes, optionally keeping
image manager untouched. All shaders, meshes, objects will be
re-allocated next time rendering happens.
Cycles cession and scene will be re-created from scratch if render/
scene parameters were changed.
This will also allow to keep compiled OSL shaders in memory without
need to re-compile them again.
P.S. Performance panel could be cleaned up a bit, not so much happy
with it's vertical alignment currently but not sure how to make
it look better.
P.P.S. Currently the only way to free images from the device is to
disable Persistent Images option and start rendering.
There were a bunch of other issues with dupli motion blur and syncing, the problem
being that there was no proper way to detect corresponding duplis between frames
or updates. As a solution, a persistent_id was added to the DupliObject. It's an
extension of the previous index value, with one index for each dupli level. This
can be used to reliably find matching dupli objects between frames. Works with
nested duplis, multiple particle systems, etc.
for now subtype is not defined, but once we start parsing the metadata we can set texture inputs as FILEPATH
also, it takes relative strings and convert to absolute for all strings (which is arguably a good solution, but
should work for now)
rays from the OSL shader. The "shade" parameter is not supported currently, but
attributes can be retrieved from the object that was hit using the
getmessage("trace", ..) function.
As mentioned in the OSL specification, this function can't be used instead of
lighting, the main purpose is to allow shaders to "probe" nearby geometry, for
example to apply a projected texture that can be blocked by geometry, apply
more “wear” to exposed geometry, or make other ambient occlusion-like effects.
http://wiki.blender.org/index.php/Doc:2.6/Manual/Render/Cycles/Nodes/OSL#Trace
Example .blend and render:
http://www.pasteall.org/blend/17347http://www.pasteall.org/pic/show.php?id=40066
That was more like workaround which solved particular issue with freestyle
rendering, but introduced some more uncontrollable issues, main of which is
recursive fracture would likely suffer from non-manifolds results of previous
step.
Boolean operation is not the only way to generate zero-area face, it could
also happen with character animation and it's indeed better be solved from
freestule side.
This should fix: #33041 Boolean difference sometimes produces meshes with holes
Issue was caused by changed order of texture slots -- float textures
have got lower slots indices than byte textures. OpenCL was still assuming
byte textures goes before float.
Now tile size is setting up explicitly instead of using number of tiles.
This allows better control over GPU performance, where having tiles aligned
to specific size makes lots of sense.
Still to come: need to update startup.blend to make tiles size 64x64.
This commit adds memory usage information while rendering.
It reports memory used by device, meaning:
- For CPU it'll report real memory consumption
- For GPU rendering it'll report GPU memory consumption, but it'll
also mean the same memory is used from host side.
This information displays information about memory requested by Cycles,
not memory really allocated on a device. Real memory usage might be
higher because of memory fragmentation or optimistic memory allocator.
There's really nothing we can do against this.
Also in contrast with blender internal's render cycles memory usage
does not include memory used by scene, only memory needed by cycles
itself will be displayed. So don't freak out if memory usage reported
by cycles would be much lower than blender internal's.
This commit also adds RenderEngine.update_memory_stats callback which
is used to tell memory consumption from external engine to blender.
This information is used to generate information line after rendering
is finished.
"interface" appeared to be a shadowed var (or even a macro/define elsewhere)
which was causing errors like:
intern\cycles\blender\blender_mesh.cpp:124:23: error: multiple types in one
declaration
intern\cycles\blender\blender_mesh.cpp:124:23: error: declaration does not
declare anything [-fpermissive]
Windows systems: text editor, text object, doesn't allow to input the standard
set of ALT+key characters (the ones coming from the OS).
- Reverted fix january 2012 - that meant to use text object ALT+keys
- Removed old (90ies) feature for hardcoded special characters in Text object.
(OS delivers this now)
Will wait for confirmation by windows compiler :)
Documentation here:
http://wiki.blender.org/index.php/Doc:2.6/Manual/Render/Cycles/Nodes/OSLhttp://wiki.blender.org/index.php/Dev:Ref/Release_Notes/2.65/Cycles
These changes require an OSL build from this repository:
https://github.com/DingTo/OpenShadingLanguage
The lib/ OSL has not been updated yet, so you might want to keep OSL disabled
until that is done.
Still todo:
* Auto update for external .osl files not working currently, press update manually
* Node could indicate better when a refresh is needed
* Attributes like UV or generated coordinates may be missing when requested from
an OSL shader, need a way to request them to be loaded by cycles
* Expose string, enum and other non-socket parameters
* Scons build support
Thanks to Thomas, Lukas and Dalai for the implementation.
* Moved kernel/osl/nodes to kernel/shaders
* Renamed standard attributes to use geom:, particle:, object: prefixes
* Update stdosl.h to properly reflect the closures we support
* Fix the wrong stdosl.h being used for building shaders
* Add geom:numpolyvertices, geom:trianglevertices, geom:polyvertices attributes
Patch by Yasuhiro Fujii, thanks!
Original issue was that in vases viewport's lens are different from default
value switching between perspective and orthographic projections will change
viewplane a lot, which is disorienting and annoying.
This time issue was caused by Save Buffers enabled and the reason blender
crashed was tiles generated by cycles didn't match tiles generated by
openexr.
Currently solved it in a way that background rendering will allocate tiles
in the same way render parts are generated in blender. Viewport rendering
will still be using sliced image, but it's more tricky to change and it's
no so big deal for now.
Issue was caused by offline rendering could have been allocated the same tile
to different devices and in this case buffers would become invalid.
Made it more clear in the code, so now it's flag for tile manager to indicate
whether tiles should always be allocated for the same device or not.
Also cleaned a way how tile index for progressive refine is calculating,
which is now avoids tricky computation based on tile coordinate and it's
dimensions.
Added support of such features, as:
- Ability to call RNA functions using C++ classes
For example RenderEngine.tag_update
- Property setters (for scalars and arrays)
Used Qt/jQuery-like getters/setters style, meaning Class.prop() is a getter,
Class.prop(value) is a setter.
Still to come:
Collection functions are not currently registering inside a property
Meaning BlendData.meshes wouldn't be a subclass of BlendDataMeshes result
you'll need to explicitly create BlendDataMeshes for now instead of doing
BlendData.meshes.remove()
- use unsigned ints only (where mixing signed/unsigned)
- turn heap_swap into an inline function, add SWAP_TVAL macro to swap values using a temp value as storage.
- added type checking SHIFT3/4 macros
also style cleanup for CTR_Map
transmission pass and filter glossy option.
The BSDF closure class is now more similar to the SVM closures, and includes
some flags and labels that are needed to properly categorize the BSDF's for
render passes. Phong closure is gone for the moment, needs to be adapated to
the new structure still.
- define array sizes for functions that take vectors.
- quiet some -Wshadow warnings.
- some copy/paste error in readfile.c made it set the same particle recalc flag twice.
1) object_fetch_transform_motion omits the per-object motion blur test (r51394), must use object_fetch_transform_motion_test.
2) KernelCamera.ndctoworld has been removed (r51402), do transform invert directly.
Background attributes are used as fallback in two cases:
1) Non-object light samples (e.g. lamp shaders)
2) Fallback if no implicit object attribute can be found
- move object_iterators.c --> view3d_iterators. (ED_object.h had to include ED_view3d.h which isn't so nice)
- move projection functions from view3d_view.c --> view3d_project.c (view3d_view was becoming a mishmash of utility functions and operators).
- some some cmake includes as system-includes.
The emitter visibility option is messy design, it makes such checks unnecessarily complicated. A better approach would be to allow non-mesh objects to carry particle data, these objects would just be invisible anyway without having to care about extra settings. However, this conflicts with the simplistic particle design of "owner is the emitter" ...
This makes it possible to do a border render inside a viewport even
when not looking through the camera.
Render border could be defined by Ctrl-B shortcut (works for both
camera render border and viewport render border).
Camera render border could still be defined using Shift-B (so no
muscule memory would be broken). Currently used a special flag of
operator to do this, otherwise you'll need to either two operators
with different poll callback or it could go into conflict with a
border zoom,
Border render of a viewport could be enabled/disabled in View
panel using "Render Border" option.
objects in the scene will also cause motion blur.
This change does come with a bit of a slow down to the CPU rendering kernel even
with motion blur disabled, due to extra overhead in handling of object matrices.
It's a few percentages on simpler scenes, not so noticeable on more complex ones.
With motion blur enabled rendering is of course also slower as would be expected,
though from testing especially GPU rendering handles it quite well.
This does not support motion blur from deforming objects yet, only translation,
scale and rotation. Deformation blur is probably for another release.
Just makes progressive refine :)
This means the whole image would be refined gradually using as much
threads as it's set in performance settings. Having enough tiles is
required to have this option working as it's expected.
Technically it's implemented by repeatedly computing next sample for
all the tiles before switching to next sample.
This works around 7-12% slower than regular tile-based rendering, so
use this option only if you really need it.
This commit also fixes progressive update of image when Save Buffers
option is enabled.
And one more thing this commit fixes is handling display buffer with
Save Buffers option enabled. If this option is enabled image buffer
wouldn't have neither byte nor float buffer until image is fully
rendered which could backfire in missing image while rendering in
cases color management cache became full.
This issue solved by allocating byte buffer for image buffer from
tile update callback.
Patch was reviewed by Brecht. He also made some minor edits to
original version to patch. Thanks, man!
Each BSDF node now has a Normal input, which can be used to set a custom normal
for the BSDF, for example if you want to have only bump on one of the layers in
a multilayer material.
The Bump node can be used to generate a normal from a scalar value, the same as
what happens when you connect a scalar value to the displacement output.
Documentation has been updated with the latest changes:
http://wiki.blender.org/index.php/Doc:2.6/Manual/Render/Cycles/Nodes
Patch by Agustin Benavidez, some implementation tweaks by me.
Documentation & Test blend files:
------------------
http://wiki.blender.org/index.php/User:MiikaH/GSoC-2012-Smoke-Simulator-Improvements
Credits:
------------------
Miika Hamalainen (MiikaH): Student / Main programmer
Daniel Genrich (Genscher): Mentor / Programmer of merged patches from Smoke2 branch
Google: For Google Summer of Code 2012
It's using the Ward BSDF currently, which has some energy loss so might be a bit
dark. More/better BSDF options can be implemented later.
Patch by Mike Farnsworth, some modifications by me. Currently it's not possible yet
to set a custom tangent, that will follow as part of per-bsdf normals patch.
* The minimize() function, which solves a least-squares problem, is
only needed for sharp remesh mode, but was being calculated for
smooth and blocks modes as well. Disabling this calculation when
it's not needed gives a big performance boost.
* OSL UI message did not show up when device type was GPU, but User Preferences were None. Also remove experimental check, more convenient for testing.
Generated and UV coordinates from the duplicator of instance instead of the
object itself.
This was used in e.g. Big Buck Bunny for texturing instanced feathers with
a UV map on the bird. Many files changed, mainly to do some refactoring to
get rid of G.rendering global in duplilist code.
ocio configuration file failed to load
This solves issues with infinite NULL-checks to prevent crashes in
such situations. Currently only happens if there's no configuration
file at all, but could be tweaked further to fallback if this file
isn't usable by blender.
still look a bit strange since the viewport can't actually render such panorama views,
so the opengl drawn scene behind the border render will not match up.