4 years agoMerge branch 'master' into temp-ghash-experiments temp-ghash-experiments
Bastien Montagne [Mon, 29 Jun 2015 14:41:00 +0000 (16:41 +0200)]
Merge branch 'master' into temp-ghash-experiments

Note that 'store hash' feature was removed for now - to complex to maintain (conflicts)
and relatively easy to re-add if we ever really want this one day.


4 years agoFix T45022: Update missing when linking objects with new depsgraph
Sergey Sharybin [Mon, 29 Jun 2015 10:58:23 +0000 (12:58 +0200)]
Fix T45022: Update missing when linking objects with new depsgraph

4 years agoFix T45156: scaling region crash
Campbell Barton [Mon, 29 Jun 2015 10:46:23 +0000 (20:46 +1000)]
Fix T45156: scaling region crash

4 years agoFix T45154: Translation binary file( for Japanese is too small
Sergey Sharybin [Mon, 29 Jun 2015 10:26:15 +0000 (12:26 +0200)]
Fix T45154: Translation binary file( for Japanese is too small

The issue was caused by some changes made to msgfmt which were needed to make
modified (cleaned-up, stripped-comments messages) working.

Unfortunately that fix was merged into the release branch, so this fix is to
be ported there as well and verified against rc1 translations.

4 years agoPartial fix T45156: scaling region crash
Campbell Barton [Mon, 29 Jun 2015 06:44:18 +0000 (16:44 +1000)]
Partial fix T45156: scaling region crash

'ar->winy' may not be initialized, making regions zoom in (past limits)
and attempt to draw very large text (~10x10k size characters), often crashing.

Fix isn't complete since it only corrects factory startup.

4 years agoCleanup: Style in for loops header.
Thomas Dinges [Sun, 28 Jun 2015 22:56:04 +0000 (00:56 +0200)]
Cleanup: Style in for loops header.

4 years agoFix T45227: Light optimization commit broke world MIS
Sergey Sharybin [Sun, 28 Jun 2015 18:47:35 +0000 (20:47 +0200)]
Fix T45227: Light optimization commit broke world MIS

4 years agoCycles: Avoid having duplication of BVH arrays during build
Sergey Sharybin [Sun, 28 Jun 2015 16:07:48 +0000 (18:07 +0200)]
Cycles: Avoid having duplication of BVH arrays during build

Previous idea behind having vector during building and array for actual storage
was needed in order to minimize amount of re-allocations happening during the
build, but it lead to double memory overhead used by those arrays at the vector
to array conversion stage.

Issue with such approach was that for BVH without spatial split size of arrays
is known in advance and it never changes, which made vector to array conversion
totally redundant.

Also after testing with several rather complex from spatial split scenes (such
as trees) it seems even conservative approach of reallocation (when we perform
re-allocation when leaf does not fit into the memory) doesn't give measurable
difference in time.

This makes it so we can switch to array, which will avoid unneeded memory
re-allocations when spatial split is disabled without harming other cases.

it's a bit difficult to measure exact benefit of this change on our production
files here, but depending on the scene it might give quite reasonable memory

4 years agoCycles: Add assert to an array at() function to be sure we don't have bad memory...
Sergey Sharybin [Sun, 28 Jun 2015 09:18:48 +0000 (11:18 +0200)]
Cycles: Add assert to an array at() function to be sure we don't have bad memory access

4 years agoCycles: Minor code style cleanup, whitesaces
Sergey Sharybin [Sun, 28 Jun 2015 08:51:49 +0000 (10:51 +0200)]
Cycles: Minor code style cleanup, whitesaces

4 years agoBGE: added clamping of angular velocity.
Sybren A. Stüvel [Tue, 23 Jun 2015 13:02:28 +0000 (15:02 +0200)]
BGE: added clamping of angular velocity.

Angular velocity clamping was missing from the BGE. It is implemented
similarly to the linear velocity clamping. It is needed to be able to
drive physical simulations of systems that have a limited rotational

Reviewed by: campbellbarton, panzergame, ton

Differential Revision:

4 years agoBGE Fix: apply velocity clamping on every physics subtick
Sybren A. Stüvel [Thu, 18 Jun 2015 12:43:30 +0000 (14:43 +0200)]
BGE Fix: apply velocity clamping on every physics subtick

This patch uses the Bullet "internal tick callback" functionality to
ensure that velocity clamping is performed after every physics update.
This makes a difference when physics subticks > 1, as in that case the
too-high velocity could have impacted the simulation.

This patch follows the examples at [1] and [2]; the latter example
also explains that the way we limit velocity in the BGE (before this
patch) is wrong.


Reviewed by: panzergame

Differential Revision:

4 years agoCleanup: remove BLI prefix from BKE funcs
Campbell Barton [Sun, 28 Jun 2015 09:09:20 +0000 (19:09 +1000)]
Cleanup: remove BLI prefix from BKE funcs

4 years agoFix T45214: BI render: maximum saturation bug in shadow pass with non-shadow lighting.
Bastien Montagne [Sun, 28 Jun 2015 08:46:42 +0000 (10:46 +0200)]
Fix T45214: BI render: maximum saturation bug in shadow pass with non-shadow lighting.

In case scene lighting would only have non-shadow light source, shadow intensity
would remain to 'pitch black'...

4 years agoTemporarily disable absolute snapping
Campbell Barton [Sun, 28 Jun 2015 00:03:55 +0000 (10:03 +1000)]
Temporarily disable absolute snapping

This really should have been finalized as a design task first,
there are too many open topics on how it should work.

4 years agoFix T45216: File Browser shows negative sizes for large files.
Bastien Montagne [Sat, 27 Jun 2015 21:39:48 +0000 (23:39 +0200)]
Fix T45216: File Browser shows negative sizes for large files.

Simply backport small part of work from asset-experiments here (using double and
adding tera-bytes unit), looks like off_t is not always 64bits even on a 64bit OS...

4 years agoCycles: Use aligned blender allocator when using guarded allocation
Sergey Sharybin [Sat, 27 Jun 2015 19:07:43 +0000 (21:07 +0200)]
Cycles: Use aligned blender allocator when using guarded allocation

This way we solve possible issues caused by regular allocator not being aware of
some classes preferring 16 bytes alignment needed for SSE to work properly. This
caused random crashes during rendering.

Now we always use aligned allocation in GuardedAllocator which shouldn't be any
measurable performance impact and the code is only used by developers after
defining special symbol, so there is no impact on release builds at all.

4 years agoOpenGL debug contexts:
Antony Riakiotakis [Sat, 27 Jun 2015 15:20:30 +0000 (17:20 +0200)]
OpenGL debug contexts:

Enable debug output in debug contexts for gl 4.3+

4 years agoCycles: Fix for wrong device enumeration in CUDA
Sergey Sharybin [Sat, 27 Jun 2015 13:11:46 +0000 (15:11 +0200)]
Cycles: Fix for wrong device enumeration in CUDA

it is the same issue as described in the previous commit, original changes
in this area were wrong and only worked on a bugger optimus driver which
simply appeared to work by co-incident and in fact used wrong device..

4 years agoCycles: Rework the way how OpenCL devices are created
Sergey Sharybin [Sat, 27 Jun 2015 13:03:51 +0000 (15:03 +0200)]
Cycles: Rework the way how OpenCL devices are created

It was annoying copy-paste happened across OpenCL device constructor, device
enumeration and split kernel checks. Now those areas are using an utility
function which returns pairs of platform and device IDs for devices which are
supported by Cycles and enumeration is happening inside that list.

This makes it so filtering is happening in a single place, so there's no need
to keep 3 different functions in sync.

This commit also fixes a bug with wrong enumeration of devices caused by recent
fixes. Those fixes were in fact wrong and only happened to appear to be working
on laptop with optimus card on Linux. Root of those issues is in fact in bad
Linux driver for optimus cards.

4 years agoCycles: Allow using custom allocators for vector class
Sergey Sharybin [Fri, 26 Jun 2015 20:36:31 +0000 (22:36 +0200)]
Cycles: Allow using custom allocators for vector class

4 years agoCycles: Ignore light which has no contribution to the scene
Sergey Sharybin [Thu, 25 Jun 2015 15:00:32 +0000 (17:00 +0200)]
Cycles: Ignore light which has no contribution to the scene

This commit makes it so light which has zero energy or doesn't has
emission shader at all is being ignored by the path tracing.

4 years agoTransform: absolute grid snapping
Campbell Barton [Sat, 27 Jun 2015 05:30:17 +0000 (15:30 +1000)]
Transform: absolute grid snapping

D910 by @donfabio with edits

New icon for menu is still TODO

4 years agoCycles standalone: add support for reading UV coordinates to the XML scene reader
Martijn Berger [Sat, 27 Jun 2015 10:05:05 +0000 (12:05 +0200)]
Cycles standalone: add support for reading UV coordinates to the XML scene reader

4 years agoCleanup: int/uint mismatch in printf...
Bastien Montagne [Sat, 27 Jun 2015 09:02:58 +0000 (11:02 +0200)]
Cleanup: int/uint mismatch in printf...

4 years agoCleanup: all params of BLI_str partition funcs can be const...
Bastien Montagne [Sat, 27 Jun 2015 09:00:47 +0000 (11:00 +0200)]
Cleanup: all params of BLI_str partition funcs can be const...

4 years agoOpenEXR cleanup: get rid of public `IMB_exr_split_token`, use BLI str helpers instead...
Bastien Montagne [Sat, 27 Jun 2015 08:24:00 +0000 (10:24 +0200)]
OpenEXR cleanup: get rid of public `IMB_exr_split_token`, use BLI str helpers instead of own cooking.

4 years agoGTests for new 'end' option of `BLI_str_partition_ex()`.
Bastien Montagne [Sat, 27 Jun 2015 08:22:56 +0000 (10:22 +0200)]
GTests for new 'end' option of `BLI_str_partition_ex()`.

4 years agoExtend `BLI_str_partition_ex`: add possibility to define a right limit to the string.
Bastien Montagne [Sat, 27 Jun 2015 08:22:29 +0000 (10:22 +0200)]
Extend `BLI_str_partition_ex`: add possibility to define a right limit to the string.

Now you can define `end` pointer as right limit of the string (allows to easily search
in substring, especially useful when searching from right).

4 years agoFix part of code in `load_image_single()` wrongly disabled when WITH_OPENEXR was...
Bastien Montagne [Sat, 27 Jun 2015 07:52:52 +0000 (09:52 +0200)]
Fix part of code in `load_image_single()` wrongly disabled when WITH_OPENEXR was disabled.

4 years agoFix crasher when loading multiview OpenEXR image.
Bastien Montagne [Sat, 27 Jun 2015 07:46:38 +0000 (09:46 +0200)]
Fix crasher when loading multiview OpenEXR image.

With multiview/multilayer OpenEXR file, `load_image_single()` will return NULL ibuf,
since it has already populated ima (with `image_create_multiview()` or
`image_create_multilayer()` calls).

Also, added some more checks before doing `IMB_ImBufFromStereo3d()`, to be sure
we do have enough slots in ibuf_arr, and we do not overwrite second ibuf either.

4 years agoFix T45204: String Splitting Function Bug: Using OpenEXR function even if compiled...
Bastien Montagne [Fri, 26 Jun 2015 17:41:51 +0000 (19:41 +0200)]
Fix T45204: String Splitting Function Bug: Using OpenEXR function even if compiled without OpenEXR.


Also, fix a potential buffer overrun here.

This should be backported to final release.

4 years agoCleanup: Fix a typo in world MIS.
Thomas Dinges [Fri, 26 Jun 2015 19:36:28 +0000 (21:36 +0200)]
Cleanup: Fix a typo in world MIS.

Found by Lukas Stockner, thanks!

4 years agoFix crash on undoing after 8690ea6
Antony Riakiotakis [Fri, 26 Jun 2015 15:50:08 +0000 (17:50 +0200)]
Fix crash on undoing after 8690ea6

Forgot to clear the mutex when reloading the scene.

4 years agoMake code compile with GPU_DEBUG
Antony Riakiotakis [Fri, 26 Jun 2015 14:18:06 +0000 (16:18 +0200)]
Make code compile with GPU_DEBUG

4 years agoFix T45199 crash when editing material nodes.
Antony Riakiotakis [Fri, 26 Jun 2015 12:47:53 +0000 (14:47 +0200)]
Fix T45199 crash when editing material nodes.

Issue is data race between preview job and GPU nodetree evaluation when
localizing the nodetree. Data race happens due to localizations doing
overrides on original nodes' new_node variable.

Solution here could probably be to use a hash for mapping of old to new
nodes but will prefer simple brute force lock for now.

4 years agoCleanup: transform center
Campbell Barton [Fri, 26 Jun 2015 06:19:39 +0000 (16:19 +1000)]
Cleanup: transform center

store global center in transform struct,
some code was calculating all the time, this is useful to keep available.

4 years agoCleanup: transform aspect
Campbell Barton [Fri, 26 Jun 2015 05:45:09 +0000 (15:45 +1000)]
Cleanup: transform aspect

Transform code had duplicate aspect checking,
now store aspect in TransInfo.aspect for reuse.

4 years agoRevert "Fix off by one error in display of start/end frame in sequencer."
Antony Riakiotakis [Thu, 25 Jun 2015 18:10:30 +0000 (20:10 +0200)]
Revert "Fix off by one error in display of start/end frame in sequencer."

This reverts commit 0e02ad8b6495b2755598a24b487041a3ed8e0116.

Initial commit was done so visual result fits with animation cursor
in timeline but this makes it so it looks like one extra frame is
rendered. Other idea would be to render one less frame for sequencer
but this is not so nice either. Generally here's no way to be
fully consistent here, but at least let's be workflow-consistent

4 years agoFix T45191 Speed strip behaviour not easy to predict.
Antony Riakiotakis [Thu, 25 Jun 2015 17:49:08 +0000 (19:49 +0200)]
Fix T45191 Speed strip behaviour not easy to predict.

Code here calculated speed based on underlying strip start position,
which was not really visible, making prediction of the result really
difficult. Things here are simple: As long as the strip exists,
manipulate the current frame by the provided factor.

4 years agoFix T45190 effect muting does not restore original sequencer display.
Antony Riakiotakis [Thu, 25 Jun 2015 16:21:58 +0000 (18:21 +0200)]
Fix T45190 effect muting does not restore original sequencer display.

4 years agoCorrect error in recent refactor
Campbell Barton [Thu, 25 Jun 2015 12:11:25 +0000 (22:11 +1000)]
Correct error in recent refactor

Closed loops missed last line

4 years agoSelect flush was missing in delete edge-loop
Campbell Barton [Thu, 25 Jun 2015 11:06:47 +0000 (21:06 +1000)]
Select flush was missing in delete edge-loop

4 years agosplash fix, previous had color conversion error
Campbell Barton [Thu, 25 Jun 2015 08:29:20 +0000 (18:29 +1000)]
splash fix, previous had color conversion error

4 years agoWeight Paint: replace Blend with Smooth tool
Campbell Barton [Thu, 25 Jun 2015 06:10:18 +0000 (16:10 +1000)]
Weight Paint: replace Blend with Smooth tool

Improved behavior

- can smooth # iterations
- option to expand/contract weights
- optionally mix with all/selected/unselected

4 years agoAPI calls for converting weights to float array
Campbell Barton [Thu, 25 Jun 2015 05:59:48 +0000 (15:59 +1000)]
API calls for converting weights to float array

4 years agoCleanup: generalize weight paint poll function
Campbell Barton [Wed, 24 Jun 2015 22:32:09 +0000 (08:32 +1000)]
Cleanup: generalize weight paint poll function

4 years agoBGE: Missing dot in 2d filter actuator documentation.
Quentin Wenger [Wed, 24 Jun 2015 20:31:11 +0000 (22:31 +0200)]
BGE: Missing dot in 2d filter actuator documentation.

4 years agoBGE: Fix 2dfilter actuator mode constants.
Quentin Wenger [Wed, 24 Jun 2015 13:31:22 +0000 (15:31 +0200)]
BGE: Fix 2dfilter actuator mode constants.

4 years agoBGE: Fix color used as background in VideoTexture.
Quentin Wenger [Wed, 24 Jun 2015 10:52:54 +0000 (12:52 +0200)]
BGE: Fix color used as background in VideoTexture.

Now we use color converted (if we do a color management) by the setter for background color in VideoTexture (ImageRender & ImageMirror).


4 years agoWeightPaint Blend: don't stack mem for dverts
Campbell Barton [Wed, 24 Jun 2015 10:42:02 +0000 (20:42 +1000)]
WeightPaint Blend: don't stack mem for dverts

Bad assumption since this could be a large list

4 years agoCleanup: warnings
Campbell Barton [Wed, 24 Jun 2015 08:42:16 +0000 (18:42 +1000)]
Cleanup: warnings

4 years agoFix edge/vert slide UV-correct, small face error
Campbell Barton [Wed, 24 Jun 2015 01:09:53 +0000 (11:09 +1000)]
Fix edge/vert slide UV-correct, small face error

Decrease epsilon to prevent flickering with small faces.

4 years agoFix edge/vert slide UV-correct & non-planar faces
Campbell Barton [Wed, 24 Jun 2015 00:59:45 +0000 (10:59 +1000)]
Fix edge/vert slide UV-correct & non-planar faces

non-flat ngons would give instability (bad UV's).

4 years agoCorrect recent error
Campbell Barton [Wed, 24 Jun 2015 00:34:38 +0000 (10:34 +1000)]
Correct recent error

4 years agoRNA: exposing image_user settings for ImageTextureNodes
Dalai Felinto [Wed, 24 Jun 2015 00:17:55 +0000 (21:17 -0300)]
RNA: exposing image_user settings for ImageTextureNodes

This is required in order to access image sequence frame_duration and
frame_offset among other settings.

4 years agoFix edge/vert slide UV-correct & zero length edges
Campbell Barton [Tue, 23 Jun 2015 23:54:23 +0000 (09:54 +1000)]
Fix edge/vert slide UV-correct & zero length edges

When calculating loop angle weighting, skip overlapping vertices.

4 years agoImprove dist_***_to_corner_v3v3v3 precision
Campbell Barton [Tue, 23 Jun 2015 21:14:33 +0000 (07:14 +1000)]
Improve dist_***_to_corner_v3v3v3 precision

Remove offset before calculating distance.

Define 'plane3' to BLI_math, since we often don't need the 4th component.

4 years agoCleanup: style
Campbell Barton [Tue, 23 Jun 2015 19:13:43 +0000 (05:13 +1000)]
Cleanup: style

4 years agoSubsurf: Make color layer aquisition order for textured draw match
Antony Riakiotakis [Tue, 23 Jun 2015 15:48:29 +0000 (17:48 +0200)]
Subsurf: Make color layer aquisition order for textured draw match

4 years agoFix T45051: Curve parent bug.
Bastien Montagne [Tue, 23 Jun 2015 15:35:55 +0000 (17:35 +0200)]
Fix T45051: Curve parent bug.

PARCURVE is deprecated parting type, should never have been exposed to user!

Not a regression, but safe enough for final 2.75 imho.

4 years agoBGE: Fix 57065a, missing dots.
Porteries Tristan [Tue, 23 Jun 2015 12:57:38 +0000 (14:57 +0200)]
BGE: Fix 57065a, missing dots.

4 years agoMinor tweaks to bge.constraints docs
Campbell Barton [Tue, 23 Jun 2015 11:58:08 +0000 (21:58 +1000)]
Minor tweaks to bge.constraints docs

4 years agoUpdate bge.constraints API doc
Campbell Barton [Tue, 23 Jun 2015 11:29:37 +0000 (21:29 +1000)]
Update bge.constraints API doc

D1357 by @Matpi with edits.

4 years agoFix T45123: 2D line intersection fails
Campbell Barton [Tue, 23 Jun 2015 10:37:53 +0000 (20:37 +1000)]
Fix T45123: 2D line intersection fails

Co-linear lines could detect as intersecting even if they weren't overlapping.

4 years agoCleanup: use 2d math funcs for line intersection
Campbell Barton [Tue, 23 Jun 2015 10:28:32 +0000 (20:28 +1000)]
Cleanup: use 2d math funcs for line intersection

4 years agoFix stupid mistake
Antony Riakiotakis [Tue, 23 Jun 2015 10:58:33 +0000 (12:58 +0200)]
Fix stupid mistake

4 years agoCode cleanup: Use enums for redraw timer operator, makes things more
Antony Riakiotakis [Tue, 23 Jun 2015 10:53:33 +0000 (12:53 +0200)]
Code cleanup: Use enums for redraw timer operator, makes things more

4 years agoFix edge drawing, total loose edges can only be determined -after- the
Antony Riakiotakis [Tue, 23 Jun 2015 10:06:36 +0000 (12:06 +0200)]
Fix edge drawing, total loose edges can only be determined -after- the
edge buffer has been setup (this is where they are counted)

4 years agoCMake: quiet warnings in GTest
Campbell Barton [Tue, 23 Jun 2015 04:33:17 +0000 (14:33 +1000)]
CMake: quiet warnings in GTest

4 years agoCMake: support multiple args to remove_cc_flag
Campbell Barton [Tue, 23 Jun 2015 04:25:24 +0000 (14:25 +1000)]
CMake: support multiple args to remove_cc_flag

4 years agoAllow editing the text editor line directly
Campbell Barton [Tue, 23 Jun 2015 00:17:00 +0000 (10:17 +1000)]
Allow editing the text editor line directly

Alternate solution for T44855

4 years agoFix T45117: Dark dupli-face objects (regression)
Campbell Barton [Mon, 22 Jun 2015 22:02:00 +0000 (08:02 +1000)]
Fix T45117: Dark dupli-face objects (regression)

4 years agoMissed changing default arg in addon_utils.disable
Campbell Barton [Mon, 22 Jun 2015 21:25:10 +0000 (07:25 +1000)]
Missed changing default arg in addon_utils.disable

4 years agoFix T44320: UV island overlap considered linked
Campbell Barton [Mon, 22 Jun 2015 21:07:52 +0000 (07:07 +1000)]
Fix T44320: UV island overlap considered linked

4 years agoFixed compilation error in editor/animation/anim_markers
Jeroen Bakker [Mon, 22 Jun 2015 18:38:25 +0000 (20:38 +0200)]
Fixed compilation error in editor/animation/anim_markers

4 years agoFix T45149: Normal Node shows a hole in its sphere with heigh scale facs
Julian Eisel [Mon, 22 Jun 2015 18:35:10 +0000 (20:35 +0200)]
Fix T45149: Normal Node shows a hole in its sphere with heigh scale facs

4 years agoFix T45034: MirrorBall rendering on wrong camera axis
Sergey Sharybin [Mon, 22 Jun 2015 18:04:46 +0000 (20:04 +0200)]
Fix T45034: MirrorBall rendering on wrong camera axis

This was a mistake in the original code from D1079.

With the current way how direction ot mirror ball works camera should look
into negative Y direction. Corrected it in the camera matrix synchronization,
so no extra calculations are needed at the render time.

That's a bit annoying, but we'd better port it to the release branch, or
otherwise we'll end up with files created with wrong camera mapping after
2.75 release.

4 years agoFix T45148, stupid own mistake, the two functions are not the same,
Antony Riakiotakis [Mon, 22 Jun 2015 17:38:30 +0000 (19:38 +0200)]
Fix T45148, stupid own mistake, the two functions are not the same,
shouldn't have collapsed them

4 years agoRevert "Lock markers now also disallows selection of markers"
Antony Riakiotakis [Mon, 22 Jun 2015 17:37:30 +0000 (19:37 +0200)]
Revert "Lock markers now also disallows selection of markers"

This reverts commit 37fd2628058f296153a21ad359fbcbc74e9141e5.

4 years agoBGE: Fix T45110, T44174, armature animations update and mirror render.
Porteries Tristan [Mon, 22 Jun 2015 14:44:16 +0000 (16:44 +0200)]
BGE: Fix T45110, T44174, armature animations update and mirror render.

Reveiwers:Moguri, Matpi, youle

4 years agoFix T45145, multiview selection fix not working for scaled matrices.
Antony Riakiotakis [Mon, 22 Jun 2015 13:55:36 +0000 (15:55 +0200)]
Fix T45145, multiview selection fix not working for scaled matrices.

Multiview code already accounts for scale, do not scale frame before
multiplying with matrix.

4 years agoFix T45136, only draw edges if there's something to draw.
Antony Riakiotakis [Mon, 22 Jun 2015 13:18:04 +0000 (15:18 +0200)]
Fix T45136, only draw edges if there's something to draw.

4 years agoCycles: Respect duplicator's object motion blur settings
Sergey Sharybin [Fri, 12 Jun 2015 11:54:17 +0000 (13:54 +0200)]
Cycles: Respect duplicator's object motion blur settings

The idea is to make it possible to control linked duplicated objects motion
blur from the scene file without need to do overrides on the linked object
settings. Currently only supported for dupligroup duplication and all now
if duplicator object has motion blur disabled then it'll be inherited into
all the duplicated objects.

There should be no regressions/changes in look of existing files because
objects do have motion blur enabled by default.

4 years agoFix T45144: Multi-value-edit ignored range
Campbell Barton [Mon, 22 Jun 2015 09:35:00 +0000 (19:35 +1000)]
Fix T45144: Multi-value-edit ignored range

4 years agoFix camera stereo logic use /w regular select
Campbell Barton [Mon, 22 Jun 2015 08:23:59 +0000 (18:23 +1000)]
Fix camera stereo logic use /w regular select

4 years agoFix T45133: Crash drawing material buttons
Campbell Barton [Mon, 22 Jun 2015 06:51:22 +0000 (16:51 +1000)]
Fix T45133: Crash drawing material buttons

4 years agoFix memory leak /w multi-drag over a single button
Campbell Barton [Sun, 21 Jun 2015 22:22:47 +0000 (08:22 +1000)]
Fix memory leak /w multi-drag over a single button

4 years agoCleanup: Get rid of some ugly magic numbers...
Bastien Montagne [Sun, 21 Jun 2015 20:37:39 +0000 (22:37 +0200)]
Cleanup: Get rid of some ugly magic numbers...

4 years agoFix T45135: More cleanup of extreme max values in operator properties.
Bastien Montagne [Sun, 21 Jun 2015 19:56:35 +0000 (21:56 +0200)]
Fix T45135: More cleanup of extreme max values in operator properties.

INT_/FLOAT_MAX are sometimes valid choices, but most of the time more
sensible values should be used here!

4 years agoFix/Cleanup possibility to type insane values in 'add' operators options.
Bastien Montagne [Sun, 21 Jun 2015 14:06:44 +0000 (16:06 +0200)]
Fix/Cleanup possibility to type insane values in 'add' operators options.

Our 'hard limit' values was too often max_int/float here, mis-typing could
lead to crash (or infinite hanging) of Blender, see e.g.

4 years agoChange defaults for planar-face tool
Campbell Barton [Sun, 21 Jun 2015 02:52:37 +0000 (12:52 +1000)]
Change defaults for planar-face tool

4 years agoCleanup: quiet warning
Campbell Barton [Sun, 21 Jun 2015 02:30:11 +0000 (12:30 +1000)]
Cleanup: quiet warning

4 years agoSupport for platforms /wo malloc_usable_size
Campbell Barton [Sun, 21 Jun 2015 02:29:20 +0000 (12:29 +1000)]
Support for platforms /wo malloc_usable_size

Was only used for stats, netbsd doesn't define this function.

4 years agoBMesh: replace BLI_array -> BLI_stack
Campbell Barton [Sat, 20 Jun 2015 23:46:12 +0000 (09:46 +1000)]
BMesh: replace BLI_array -> BLI_stack

4 years agoBMesh: replace BLI_array -> BLI_stack
Campbell Barton [Sat, 20 Jun 2015 23:15:09 +0000 (09:15 +1000)]
BMesh: replace BLI_array -> BLI_stack

Also use more direct custom-data access.

4 years agoThreads: Cache result of syscall when querying number of system threads
Sergey Sharybin [Sat, 20 Jun 2015 20:10:30 +0000 (22:10 +0200)]
Threads: Cache result of syscall when querying number of system threads

Number of system threads is quite difficult to change without need of blender
restart, so we can cache result of the systcalls (which are not really cheap)
in order to be able to call BLI_system_thread_count() without worrying of
performance issues in that function.

Reviewers: campbellbarton

Differential Revision:

4 years agoCycles: Another fix for OSX, sm_50 experimental actually also fails to compile
Sergey Sharybin [Sat, 20 Jun 2015 17:40:23 +0000 (19:40 +0200)]
Cycles: Another fix for OSX, sm_50 experimental actually also fails to compile

Didn't notice it originally because compilation was threaded.

4 years agoCycles: Tweak to previous commit, experimental sm_52 works on Linux but not OSX
Sergey Sharybin [Sat, 20 Jun 2015 17:01:24 +0000 (19:01 +0200)]
Cycles: Tweak to previous commit, experimental sm_52 works on Linux but not OSX

4 years agoCycles: Un-inline triangle_intersect_precalc() on Apple OpenCL
Sergey Sharybin [Sat, 20 Jun 2015 16:00:30 +0000 (18:00 +0200)]
Cycles: Un-inline triangle_intersect_precalc() on Apple OpenCL

This gives quite the same problems as experimental CUDA kernels
and for until it's found a root cause of the problem we'd just
explicitly uninline the function.