blender-staging.git
5 years agoCorrection for wrong ifdef in creator
Sergey Sharybin [Mon, 6 Oct 2014 12:54:35 +0000 (14:54 +0200)]
Correction for wrong ifdef in creator

5 years agoFix T42075: DnD Material not updating render view
Campbell Barton [Mon, 6 Oct 2014 12:04:55 +0000 (14:04 +0200)]
Fix T42075: DnD Material not updating render view

5 years agoCleanup: remove old preview code
Campbell Barton [Mon, 6 Oct 2014 10:36:50 +0000 (12:36 +0200)]
Cleanup: remove old preview code

5 years agoCycles: Fix for camera-in-volume detection
Sergey Sharybin [Mon, 6 Oct 2014 10:31:31 +0000 (12:31 +0200)]
Cycles: Fix for camera-in-volume detection

Ray actually should have infinite length, so we can detect camera in a volume
which is bigger that the far clipping of the camera.

This might also give some speedup (wouldn't expect much tho) because we don't
need to re-calculate ray direction and length after every bounce now.

5 years agoCycles: Optimize of volume stack update when sampling SSS
Sergey Sharybin [Mon, 6 Oct 2014 10:23:16 +0000 (12:23 +0200)]
Cycles: Optimize of volume stack update when sampling SSS

basically we skip all non-volume objects now in the volume stack function.
Depending on the show it might give some percent of speedup.

Most of the speedup would be gained in the scenes when having SSS object
intersecting the volume and taking a reasonable amount of frame space.

5 years agoCycles: Fix for viewport rendering with debug enabled
Sergey Sharybin [Mon, 6 Oct 2014 09:52:22 +0000 (11:52 +0200)]
Cycles: Fix for viewport rendering with debug enabled

5 years agoCycles: Workaround dead-slow expf() on 64bit linux
Sergey Sharybin [Mon, 6 Oct 2014 07:43:23 +0000 (13:43 +0600)]
Cycles: Workaround dead-slow expf() on 64bit linux

Single precision exponent on 64bit linux tends to be order of magnitude slower
than double precision version even with single<->double precision conversion.

Some feedback in the mailing lists also suggests that logf() is also slow, but
this i didn't confirm here in the studio yet.

Depending on the shader setup it gives ~3% with the secret agent shot and up to
around 15% with the bmw scene here.

5 years agoCycles: Move system headers include to the top of the files
Sergey Sharybin [Mon, 6 Oct 2014 07:39:19 +0000 (13:39 +0600)]
Cycles: Move system headers include to the top of the files

This is a good practice to do anyway, plus it'll help with the upcoming change.

5 years agoCleanup: remove paranoid NULL checks
Campbell Barton [Mon, 6 Oct 2014 10:23:47 +0000 (12:23 +0200)]
Cleanup: remove paranoid NULL checks

5 years agoDyntopo: minor speedup collapsing edges
Campbell Barton [Mon, 6 Oct 2014 10:16:06 +0000 (12:16 +0200)]
Dyntopo: minor speedup collapsing edges

5 years agoFix T42093: segfault when trying to free unallocated depths zbuffer (border-zoom...
Bastien Montagne [Mon, 6 Oct 2014 08:19:47 +0000 (10:19 +0200)]
Fix T42093: segfault when trying to free unallocated depths zbuffer (border-zoom in 3DView).

Thanks to jensverwiebe (jens verwiebe) for investigations too!

5 years agoQuiet warning on 32bit system
Campbell Barton [Mon, 6 Oct 2014 07:50:20 +0000 (09:50 +0200)]
Quiet warning on 32bit system

5 years agoFix T41983: Array Modifier "Merge" bug?
Bastien Montagne [Sun, 5 Oct 2014 21:10:44 +0000 (23:10 +0200)]
Fix T41983: Array Modifier "Merge" bug?

We messed up previous fix, in 'simplified' translated merge case... :(

5 years agoCycles: Center Tile order had a slight offset to the left.
Greg Zaal [Sun, 5 Oct 2014 16:35:49 +0000 (18:35 +0200)]
Cycles: Center Tile order had a slight offset to the left.

Signed-off-by: Thomas Dinges
5 years agoFix T42081, OpenCL supports float3 since the 1.1 specification, not sure why we neede...
Thomas Dinges [Sun, 5 Oct 2014 16:10:42 +0000 (18:10 +0200)]
Fix T42081, OpenCL supports float3 since the 1.1 specification, not sure why we needed this.

5 years agoCycles: Forgot to set WITH_CYCLES_DEBUG for OSL kernel
Sergey Sharybin [Sun, 5 Oct 2014 15:43:27 +0000 (17:43 +0200)]
Cycles: Forgot to set WITH_CYCLES_DEBUG for OSL kernel

5 years agoCycles: __KERNEL_DEBUG__ wasn't set for compile-time kernels
Sergey Sharybin [Sun, 5 Oct 2014 15:42:04 +0000 (21:42 +0600)]
Cycles: __KERNEL_DEBUG__ wasn't set for compile-time kernels

5 years agoFix tgz script (included all files in submodules)
Campbell Barton [Sun, 5 Oct 2014 15:18:20 +0000 (17:18 +0200)]
Fix tgz script (included all files in submodules)

5 years agoCleanup: sculpt float -> bool
Campbell Barton [Sun, 5 Oct 2014 14:06:19 +0000 (16:06 +0200)]
Cleanup: sculpt float -> bool

5 years agoOSX/Ghost: little code cleanup
Jens Verwiebe [Sun, 5 Oct 2014 10:45:06 +0000 (12:45 +0200)]
OSX/Ghost: little code cleanup

5 years agoCycles: Make OpenCL folks happy to use __KERNEL_DEBUG__
Sergey Sharybin [Sun, 5 Oct 2014 10:00:23 +0000 (16:00 +0600)]
Cycles: Make OpenCL folks happy to use __KERNEL_DEBUG__

Quite straightforward change, the only annoying thing is that we can't use
indentation for include directive just because of the way headers inlineing
works for OpenCL.

Might do smarter job in path_source_replace_includes() but don't want to
spend time on this yet.

5 years agoCycles: Fix for debug kernel not working with CUDA
Sergey Sharybin [Sun, 5 Oct 2014 09:29:26 +0000 (15:29 +0600)]
Cycles: Fix for debug kernel not working with CUDA

5 years agoBLI_buffer: simplify buffer resize
Campbell Barton [Sun, 5 Oct 2014 08:31:47 +0000 (10:31 +0200)]
BLI_buffer: simplify buffer resize

5 years agoCycles / CUDA: Workaround to make sm_52 (Maxwell) cards work.
Thomas Dinges [Sun, 5 Oct 2014 02:13:40 +0000 (04:13 +0200)]
Cycles / CUDA: Workaround to make sm_52 (Maxwell) cards work.

* sm_52 can run a sm_50 kernel, so tell runtime detection to use that until we build a dedicated sm_52 kernel.

5 years agoCycles / CUDA: Change inline rules for BVH intersection functions.
Thomas Dinges [Sun, 5 Oct 2014 01:53:51 +0000 (03:53 +0200)]
Cycles / CUDA: Change inline rules for BVH intersection functions.

* On sm_30 and above there is no change (was not inlined already before), this just fixes a speed regression from yesterday. 6359c36ba407
* On sm_2x (tested with sm_21), I get a nice 8% speedup in the bmw scene with this. As a bonus, cubin compilation time and memory usage is significantly reduced. Regular cubin size went from 2.5MB to 2.0MB, Experimental one from 3.8MB to 2.5MB.

5 years agoPrevious scons fix is just wrong, there is a reason why lines were tried to be indented
Sergey Sharybin [Sat, 4 Oct 2014 14:31:47 +0000 (16:31 +0200)]
Previous scons fix is just wrong, there is a reason why lines were tried to be indented

5 years agoCycles: Fix wrong ifdef check around shadows record all
Sergey Sharybin [Fri, 3 Oct 2014 17:24:24 +0000 (19:24 +0200)]
Cycles: Fix wrong ifdef check around shadows record all

5 years agoFix bad RNA enum from the previous commit
Sergey Sharybin [Sat, 4 Oct 2014 14:20:14 +0000 (20:20 +0600)]
Fix bad RNA enum from the previous commit

5 years agoAnother scons fix, wrong indentation.
Thomas Dinges [Sat, 4 Oct 2014 14:07:39 +0000 (16:07 +0200)]
Another scons fix, wrong indentation.

5 years agoTypo fix for new WITH_BF_CYCLES_DEBUG option.
Thomas Dinges [Sat, 4 Oct 2014 14:01:09 +0000 (16:01 +0200)]
Typo fix for new WITH_BF_CYCLES_DEBUG option.

5 years agoFix off by one error drawing passepartout
Campbell Barton [Sat, 4 Oct 2014 13:40:41 +0000 (15:40 +0200)]
Fix off by one error drawing passepartout

5 years agoCycles: Add support for debug passes
Sergey Sharybin [Sat, 4 Oct 2014 13:00:26 +0000 (19:00 +0600)]
Cycles: Add support for debug passes

Currently only summed number of traversal steps and intersections used by the
camera ray intersection pass is implemented, but in the future we will support
more debug passes which would help checking what things makes the scene slow.
Example of such extra passes could be number of bounces, time spent on the
shader tree evaluation and so.

Implementation from the Cycles side is pretty much straightforward, could only
mention here that it's a build-time option disabled by default.

From the blender side it's implemented as a PASS_DEBUG with several subtypes
possible. This way we don't need to create an extra DNA pass type for each of
the debug passes, saving us a bits.

Reviewers: campbellbarton

Reviewed By: campbellbarton

Differential Revision: https://developer.blender.org/D813

5 years agoFix T42065: Shading issue using Array modifier
Bastien Montagne [Sat, 4 Oct 2014 10:52:06 +0000 (12:52 +0200)]
Fix T42065: Shading issue using Array modifier

Gah... Dirty normal flag should be set on *real* final dm, after merge is done. :/

5 years ago"Delete texture paint slot" told the user that it would add a texture paint slot.
Thomas Beck [Sat, 4 Oct 2014 10:21:11 +0000 (12:21 +0200)]
"Delete texture paint slot" told the user that it would add a texture paint slot.

5 years agoOSX: simplyfy last commit for fs_menu system_bookmarks
Jens Verwiebe [Sat, 4 Oct 2014 08:47:53 +0000 (10:47 +0200)]
OSX: simplyfy last commit for fs_menu system_bookmarks

5 years agoCycles: Remove a workaround for Titan GPUs, not needed anymore with the latest CUDA...
Thomas Dinges [Fri, 3 Oct 2014 23:28:46 +0000 (01:28 +0200)]
Cycles: Remove a workaround for Titan GPUs, not needed anymore with the latest CUDA compiler.

5 years agoOSX: Change the FS_CATEGORY_SYSTEM_BOOKMARKS aka "Favorites", so that the orderwise...
Jens Verwiebe [Fri, 3 Oct 2014 19:01:48 +0000 (21:01 +0200)]
OSX: Change the FS_CATEGORY_SYSTEM_BOOKMARKS aka "Favorites", so that the orderwise in sync with finder

5 years agoCycles, some tweaks to scene_intersect_shadow_all()
Thomas Dinges [Fri, 3 Oct 2014 18:41:38 +0000 (20:41 +0200)]
Cycles, some tweaks to scene_intersect_shadow_all()

* Function returns a bool, not an uint.
* Remove GPU ifdefs, this is CPU only due to malloc / qsort.

5 years agoFix for tgz script (whitespace broke it)
Campbell Barton [Fri, 3 Oct 2014 17:41:46 +0000 (19:41 +0200)]
Fix for tgz script (whitespace broke it)

5 years agoInstall_deps.sh: missing `--skip-openexr` in arg parser's parameters.
Bastien Montagne [Fri, 3 Oct 2014 16:34:09 +0000 (18:34 +0200)]
Install_deps.sh: missing `--skip-openexr` in arg parser's parameters.

Reported through mail by Jefferson Rausseo, thanks!

5 years agoCleanup: Remove some unused / unreferenced functions for perdiodic perlin noise.
Thomas Dinges [Fri, 3 Oct 2014 16:00:45 +0000 (18:00 +0200)]
Cleanup: Remove some unused / unreferenced functions for perdiodic perlin noise.

5 years agoCycles / OSL: Make the signed/unsigned Perlin parameter more self explaining.
Thomas Dinges [Fri, 3 Oct 2014 15:47:37 +0000 (17:47 +0200)]
Cycles / OSL: Make the signed/unsigned Perlin parameter more self explaining.

5 years agoCycles: Fix OpenCL compile after new Volume BVH introduction and add some comments.
Thomas Dinges [Fri, 3 Oct 2014 15:23:45 +0000 (17:23 +0200)]
Cycles: Fix OpenCL compile after new Volume BVH introduction and add some comments.

5 years agoCycles: Missing volume traversal header in cmake for GPU compilation.
Thomas Dinges [Fri, 3 Oct 2014 15:11:00 +0000 (17:11 +0200)]
Cycles: Missing volume traversal header in cmake for GPU compilation.

5 years agoCycles: Remove Westin closure.
Thomas Dinges [Fri, 3 Oct 2014 14:03:49 +0000 (16:03 +0200)]
Cycles: Remove Westin closure.

Was hooked up last year for testing purposes, as we already had some code for it, but the closure itself is not really good nor really useful, so let's remove it.

5 years ago2.73 Release Cycle: Alpha.
Thomas Dinges [Fri, 3 Oct 2014 13:43:30 +0000 (15:43 +0200)]
2.73 Release Cycle: Alpha.

5 years agoCleanup: Spelling.
Thomas Dinges [Fri, 3 Oct 2014 13:28:52 +0000 (15:28 +0200)]
Cleanup: Spelling.

5 years agoFix T41983: Array regression with center-verts
Campbell Barton [Fri, 3 Oct 2014 12:33:50 +0000 (14:33 +0200)]
Fix T41983: Array regression with center-verts

Array with rotation and a central pivot would fail.

Thanks to Bastien Montagne for the initial fix.

5 years agoFix for 2-sided faces in array modifier
Campbell Barton [Fri, 3 Oct 2014 11:59:06 +0000 (13:59 +0200)]
Fix for 2-sided faces in array modifier

5 years agoCycles: Use more accurate implementation of erf() and erfinv()
Sergey Sharybin [Fri, 3 Oct 2014 12:27:08 +0000 (18:27 +0600)]
Cycles: Use more accurate implementation of erf() and erfinv()

This functions are  orders of magnitude more accurate than the old ones,
and they're around the same complexity to compute.

5 years agoSculpting:
Antony Riakiotakis [Fri, 3 Oct 2014 11:14:10 +0000 (13:14 +0200)]
Sculpting:

Grab and Snake hook brushes now support strength. To keep compatibility
with older brushes, this commit also sets strength to 1.0 for those
brushes.

5 years agoCycles: Decouple object flags update to a separate update step
Sergey Sharybin [Fri, 3 Oct 2014 10:11:19 +0000 (12:11 +0200)]
Cycles: Decouple object flags update to a separate update step

This way there's much less cross-references between objects and meshes
device update functions.

The only thing remained s the object bounds calculation which is needed
by bvh update. This could also be decoupled, but it's not that crucial
yet because its's how it used to be for ages now.

5 years agoBetter fix for T42054 - modifiers have a callback to say whether they need normals...
Bastien Montagne [Fri, 3 Oct 2014 10:12:10 +0000 (12:12 +0200)]
Better fix for T42054 - modifiers have a callback to say whether they need normals or no.

Thanks Campbell for the headup!

5 years agoFix T42054: Problem when using Cast modifier with Wireframe modifier.
Bastien Montagne [Fri, 3 Oct 2014 09:58:29 +0000 (11:58 +0200)]
Fix T42054: Problem when using Cast modifier with Wireframe modifier.

Wireframe modifier needs valid normals, let's ensure that!

5 years agoFix T41920: Changing Use Alpha settings doesn't refresh viewport properly
Sergey Sharybin [Fri, 3 Oct 2014 09:27:05 +0000 (11:27 +0200)]
Fix T41920: Changing Use Alpha settings doesn't refresh viewport properly

5 years agoCycles: Implement preliminary test for volume stack update from SSS
Sergey Sharybin [Fri, 3 Oct 2014 08:52:04 +0000 (10:52 +0200)]
Cycles: Implement preliminary test for volume stack update from SSS

This adds an AABB collision check for objects with volumes and if there's a
collision detected then the object will have SD_OBJECT_INTERSECTS_VOLUME flag.

This solves a speed regression introduced by the fix for T39823 by skipping
volume stack update in cases no volumes intersects the current SSS object.

5 years agoCycles: Fix for non-initialized variable
Sergey Sharybin [Fri, 3 Oct 2014 08:44:24 +0000 (10:44 +0200)]
Cycles: Fix for non-initialized variable

5 years agoCycles: Make camera-in-volume an official feature
Sergey Sharybin [Thu, 2 Oct 2014 21:46:31 +0000 (03:46 +0600)]
Cycles: Make camera-in-volume an official feature

This means it's no longer needed to enable experimental feature set in order to
have proper camera in volume support. And this also means if there's something
wrong going on, or if there's speed regression for cases when camera is obviously
not in the volume -- this issues are to be reported and handled in the regular
matter.

Happy blending!

5 years agoCycles: Speedup of kernel side camera-in-volume detection
Sergey Sharybin [Thu, 2 Oct 2014 21:42:16 +0000 (03:42 +0600)]
Cycles: Speedup of kernel side camera-in-volume detection

The idea is to only count intersections with objects which has volumetric shader
and ignore all other objects.

This is probably as fast as we can go without involving some forth level magic.

5 years agoCleanup: spelling, ws
Campbell Barton [Fri, 3 Oct 2014 06:09:00 +0000 (08:09 +0200)]
Cleanup: spelling, ws

5 years agoComments: note on bmesh_region_match
Campbell Barton [Fri, 3 Oct 2014 06:10:02 +0000 (08:10 +0200)]
Comments: note on bmesh_region_match

also correct buffer type

5 years agoUpdate link to the Support an Animation Movie link
Sergey Sharybin [Thu, 2 Oct 2014 18:47:24 +0000 (20:47 +0200)]
Update link to the Support an Animation Movie link

This is to be in the final release.

5 years agoCycles: optimization of panoramic camera in volume
Sergey Sharybin [Thu, 2 Oct 2014 18:37:05 +0000 (20:37 +0200)]
Cycles: optimization of panoramic camera in volume

Now we do much better preliminary check for panoramic camera is inside the
volume object boundings.

Also we're now cacheing the has_volume in the mesh, which makes it unneeded
iterations for each object's shaders.

Should be no functional changes, just faster sync and panoramic-in-volume
rendering.

5 years agoCode cleanup:
Antony Riakiotakis [Thu, 2 Oct 2014 18:18:42 +0000 (20:18 +0200)]
Code cleanup:

calc_manipulator_stats is only used in manipulator (and soon will only
use the context for such queries)

5 years agoFix for wrong bit mask magic happening in gpu_update_lamps_shadows()
Sergey Sharybin [Thu, 2 Oct 2014 16:19:54 +0000 (18:19 +0200)]
Fix for wrong bit mask magic happening in gpu_update_lamps_shadows()

5 years agoFix black matcap when using particle system
Sergey Sharybin [Thu, 2 Oct 2014 16:17:33 +0000 (18:17 +0200)]
Fix black matcap when using particle system

Using matcap with a displist is not really good thing to try
and it's actually a regression since we've enabled matcaps
for all objects instead of just an active one.

5 years agoLibmv: Fix wrong variable use in scons
Sergey Sharybin [Thu, 2 Oct 2014 14:51:58 +0000 (16:51 +0200)]
Libmv: Fix wrong variable use in scons

5 years agoFix error in last commit
Campbell Barton [Thu, 2 Oct 2014 15:10:24 +0000 (17:10 +0200)]
Fix error in last commit

5 years agoFix T42049: Crash exiting /w GL1.1
Campbell Barton [Thu, 2 Oct 2014 14:21:26 +0000 (16:21 +0200)]
Fix T42049: Crash exiting /w GL1.1

5 years agoFix T42030: Grabbing the whole mask interfere with grabbing individual curve
Sergey Sharybin [Thu, 2 Oct 2014 13:28:07 +0000 (19:28 +0600)]
Fix T42030: Grabbing the whole mask interfere with grabbing individual curve

This fix is for the final 2.72 release.

5 years agoFreestyle: For until c++11 is accepted only static members could b initialized in...
Sergey Sharybin [Thu, 2 Oct 2014 13:24:16 +0000 (19:24 +0600)]
Freestyle: For until c++11 is accepted only static members could b initialized in-lined

5 years agoFix T42008: Dragging and packing issue about new blank image
Sergey Sharybin [Thu, 2 Oct 2014 13:04:38 +0000 (19:04 +0600)]
Fix T42008: Dragging and packing issue about new blank image

The issue was caused by the fact that we never used to store the
generated image color in DNA, so image reload will loose this
information.

Now we store the color in DNA, making ti so re-loading the image
will preserve it's generated color.

It is now also possible to change generated image color using the
color swatch in image properties after the image was created.

5 years agoFix T42033 UVs shown while rendering in texture paint mode.
Antony Riakiotakis [Thu, 2 Oct 2014 12:34:53 +0000 (14:34 +0200)]
Fix T42033 UVs shown while rendering in texture paint mode.

This should be included in final release build.

5 years agoFix for regression on bevel material: need do_version support for earlier files.
Howard Trickey [Thu, 2 Oct 2014 11:49:11 +0000 (07:49 -0400)]
Fix for regression on bevel material: need do_version support for earlier files.

5 years agoFix freestyle/bplayer build (usual stub stuff).
Bastien Montagne [Thu, 2 Oct 2014 10:47:05 +0000 (12:47 +0200)]
Fix freestyle/bplayer build (usual stub stuff).

5 years agoFreestyle: Fix for a compiler failure in makesrna.
Tamito Kajiyama [Thu, 2 Oct 2014 08:49:35 +0000 (17:49 +0900)]
Freestyle: Fix for a compiler failure in makesrna.

5 years agoFreestyle: Fix for destructive view map modifications during chaining operations.
Tamito Kajiyama [Thu, 2 Oct 2014 06:26:53 +0000 (15:26 +0900)]
Freestyle: Fix for destructive view map modifications during chaining operations.

The view map is mostly treated as a read-only data structure by line stylization
operations (i.e., selection, chaining, splitting, sorting and stroke creation).  The
only exception is the chaining operation in some cases where insertion of extra
FEdge objects is necessary to ensure the continuity of underlying FEdges from
which a chain is constructed.

The present revision addresses the removal of extra FEdges so to keep the view
map clean and suitable for reuse in subsequent render frames.

5 years agoFreestyle: Finer memory statistics with CXX_GUARDEDALLOC.
Tamito Kajiyama [Thu, 2 Oct 2014 06:18:10 +0000 (15:18 +0900)]
Freestyle: Finer memory statistics with CXX_GUARDEDALLOC.

5 years agoFreestyle: View map caching.
Tamito Kajiyama [Tue, 28 Jan 2014 14:24:59 +0000 (23:24 +0900)]
Freestyle: View map caching.

New render layer option named "View map cache" is added to reuse a
previously computed view map for subsequent rendering.  The cache is
automatically updated when the mesh geometry of the input 3D scene has
been changed.

This functionality offers a major performance boost for Freestyle
animation rendering when camera-space mesh geometry is static, as well
as for repeated still renders with updates of line stylization options.

Although the "View map cache" toggle is a render layer option, the cache
memory is shared by all render layers and scenes.  This means that if
Freestyle is used for two or more render layers (possibly in different
scenes through the compositor), then the cached view map for one render
layer is replaced by a new view map for another render layer and hence
no performance gain is expected.

5 years agoCleanup: indentation
Campbell Barton [Thu, 2 Oct 2014 08:32:59 +0000 (10:32 +0200)]
Cleanup: indentation

5 years agoCompositor: Reduce number of int<->float conversions in keying screen operation
Sergey Sharybin [Wed, 1 Oct 2014 10:44:35 +0000 (16:44 +0600)]
Compositor: Reduce number of int<->float conversions in keying screen operation

5 years agoCompositor: Simplify logic in keying blur operation
Sergey Sharybin [Wed, 1 Oct 2014 10:36:28 +0000 (16:36 +0600)]
Compositor: Simplify logic in keying blur operation

5 years agoLibmv: Remove old workaround for 10.5 hash tables
Sergey Sharybin [Wed, 1 Oct 2014 06:01:24 +0000 (12:01 +0600)]
Libmv: Remove old workaround for 10.5 hash tables

We on'don't really support 10.5 anymore for my knowledge and in any case
the defined symbol was never checked in the sources, so this workaround
was basically doing nothing for quite some time now.

5 years agoLibmv: Add an option to disable schur specializations build-time
Sergey Sharybin [Wed, 1 Oct 2014 05:59:43 +0000 (11:59 +0600)]
Libmv: Add an option to disable schur specializations build-time

This way folks with few ram or those who wants to hack into Ceres
code might compile Blender much faster.

5 years agoLibmv: Silence some warnings about C11 usage
Sergey Sharybin [Wed, 1 Oct 2014 05:48:24 +0000 (11:48 +0600)]
Libmv: Silence some warnings about C11 usage

This makes headers a bit less clean (with the anonymous structs
mainly and headers inclusion from another header.

Makes compilation with clang a bit less noisy.

5 years agofix T42046 Improved the option label for clarity
Gaia Clary [Thu, 2 Oct 2014 06:54:59 +0000 (08:54 +0200)]
fix T42046 Improved the option label for clarity

5 years agoCode cleanup, use bit shifting for lamp flags.
Antony Riakiotakis [Wed, 1 Oct 2014 17:16:06 +0000 (19:16 +0200)]
Code cleanup, use bit shifting for lamp flags.

5 years agoFix T41971: BMesh mapping slot arg error
Campbell Barton [Wed, 1 Oct 2014 12:36:09 +0000 (22:36 +1000)]
Fix T41971: BMesh mapping slot arg error

5 years agoAnim Playback: Optionally draw flipped
Campbell Barton [Wed, 1 Oct 2014 11:59:02 +0000 (21:59 +1000)]
Anim Playback: Optionally draw flipped

useful to view animations from different POV.
FKey: X, Shift+FKey: Y

5 years agoCleanup: de-duplicate playanim matrix
Campbell Barton [Wed, 1 Oct 2014 10:45:05 +0000 (20:45 +1000)]
Cleanup: de-duplicate playanim matrix

5 years agoFix Smoke UI property: Low resolution values are needed for pyroclastic smoke effects.
Daniel Genrich [Wed, 1 Oct 2014 11:32:52 +0000 (13:32 +0200)]
Fix Smoke UI property: Low resolution values are needed for pyroclastic smoke effects.
Example values for such an effects are resolution = 10 and high resolution = 5.

Patch by nudelZ

5 years agoSmooth vertex: User request: tweak hard min/max smoothness.
Bastien Montagne [Wed, 1 Oct 2014 08:46:50 +0000 (10:46 +0200)]
Smooth vertex: User request: tweak hard min/max smoothness.

Keep soft min/max to usual values, but allow artists to play with strange values
if they really want it.

5 years agoFreestyle: New line style options for sorting and chain selection.
Tamito Kajiyama [Wed, 1 Oct 2014 06:42:37 +0000 (15:42 +0900)]
Freestyle: New line style options for sorting and chain selection.

The following two sort keys are added for sorting chains.

* Projected X - Sort by the projected X value in the image coordinate system.
* Projected Y - Sort by the projected Y value in the image coordinate system.

A new line style option for the selection of first N chains is also added.

Moreover, the chain sorting and chain selection operations are now executed
in this order instead of the reverse order used previously.  The UI has also
changed accordingly.  This functional change is backward compatible and
won't result in visual differences.

5 years agoFreestyle: Fix for a typo in the definition of NotBP1D.
Tamito Kajiyama [Wed, 1 Oct 2014 06:32:46 +0000 (15:32 +0900)]
Freestyle: Fix for a typo in the definition of NotBP1D.

This is a safe fix for a regression from 2.71, so should be back ported to 2.72.

5 years agoCycles: Filter Shader and Output nodes to their respective object/world node tree.
Thomas Dinges [Wed, 1 Oct 2014 04:23:43 +0000 (06:23 +0200)]
Cycles: Filter Shader and Output nodes to their respective object/world node tree.

Some nodes only work in certain node trees, so don't show them in the Add Node menu when this is the case.
This can probably be expanded to Input Nodes too, but need to double check some cases here still.

5 years agoFix for duplicate member in UserDef from own commit (merge mistake).
Lukas Tönne [Tue, 30 Sep 2014 16:37:42 +0000 (18:37 +0200)]
Fix for duplicate member in UserDef from own commit (merge mistake).

5 years agoSupport for "virtual pixelsize" on 4k screens.
Lukas Tönne [Tue, 30 Sep 2014 16:29:32 +0000 (18:29 +0200)]
Support for "virtual pixelsize" on 4k screens.

On 4k devices the default pixel size leads to tiny OpenGL drawing
that is hardly usable without doubling the DPI. The retina system
on OSX aims to alleviate this problem by introducing a general 2x
pixel size.

No equivalent feature exists on other platforms so far. However,
to emulate the effect this patch introduces a "virtual" pixel size
factor for OpenGL drawing.

Note that the user currently has to enable this manually by selecting
the "Virtual Pixel Mode" in the user preferences (defaults to native).
All windows of a Blender instance share the same virtual pixel size as well.
It may be possible to handle this on a per-window basis and automate
the selection somewhat (if enabled by the user), so working with
multiple screens becomes more convenient, but technical limitations
make this a bit difficult (on X11 with nvidia drivers the actual screen size
is not reported correctly).

Reviewers: ton, campbellbarton

Differential Revision: https://developer.blender.org/D669

5 years agoFix T42026: Unit scale affects camera focal length.
Bastien Montagne [Tue, 30 Sep 2014 13:07:44 +0000 (15:07 +0200)]
Fix T42026: Unit scale affects camera focal length.

Not a regression, yet maybe simple/safe enough for 2.72?

5 years agoFix compilation on GCC (good old mixed declaration and code).
Bastien Montagne [Tue, 30 Sep 2014 12:22:20 +0000 (14:22 +0200)]
Fix compilation on GCC (good old mixed declaration and code).

5 years agoStyle cleanup
Sergey Sharybin [Tue, 30 Sep 2014 11:28:04 +0000 (17:28 +0600)]
Style cleanup