blender.git
5 years agoMerged revision(s) 58859-58993 from trunk/blender into soc-2013-dingto.
Thomas Dinges [Wed, 7 Aug 2013 16:57:18 +0000 (16:57 +0000)]
Merged revision(s) 58859-58993 from trunk/blender into soc-2013-dingto.

5 years agoFix #36373: maya and max key configurations for mesh loop select did not work correct
Brecht Van Lommel [Wed, 7 Aug 2013 15:51:57 +0000 (15:51 +0000)]
Fix #36373: maya and max key configurations for mesh loop select did not work correct
when doing an extend loop select, then doing a regular loop select, it would still
extend.

5 years agoFix #36391: removing an armature with a custom bone shape object would not
Brecht Van Lommel [Wed, 7 Aug 2013 15:23:09 +0000 (15:23 +0000)]
Fix #36391: removing an armature with a custom bone shape object would not
decrement the object user count when removing the armature. This could cause
the object to stick when it shouldn't, in particular when that object is part
of a group.

5 years agoFix #36266, First undo in compositor decrements usercounter without a reason. The...
Lukas Toenne [Wed, 7 Aug 2013 13:14:50 +0000 (13:14 +0000)]
Fix #36266, First undo in compositor decrements usercounter without a reason. The handling of scene backpointers in render layer/composite nodes changed slightly recently, which caused a double increment
of the scene user count. The node->id pointer for these nodes is now initialized in the respective init callbacks already. The explicit assignment and increment in the ED_node_composit_default is not
necessary and just adds an increment without checking previous values and decrementing them properly.

Note that the current system still leaves the scene with "fake" users (rlayer + composite nodes) which are actually part of the scene data itself. But that's design issue with the "local" node tree data
and doesn't do any real harm.

5 years agoremove assert in check_for_dupid() function for rare but valid renaming situation.
Campbell Barton [Wed, 7 Aug 2013 08:20:37 +0000 (08:20 +0000)]
remove assert in check_for_dupid() function for rare but valid renaming situation.

5 years agoAdding a fov attribute to KX_Camera. This attribute converts the field of view value...
Daniel Stokes [Wed, 7 Aug 2013 05:24:47 +0000 (05:24 +0000)]
Adding a fov attribute to KX_Camera. This attribute converts the field of view value and uses it to set lens.

5 years agoquiet shadow warnings and remove redundant NULL check
Campbell Barton [Wed, 7 Aug 2013 04:31:05 +0000 (04:31 +0000)]
quiet shadow warnings and remove redundant NULL check

5 years agocode cleanup: more confusion with 0/NULL/false
Campbell Barton [Wed, 7 Aug 2013 03:55:21 +0000 (03:55 +0000)]
code cleanup: more confusion with 0/NULL/false

5 years agocode cleanup: some structs were declaring data when only typedef's were intended...
Campbell Barton [Wed, 7 Aug 2013 03:44:05 +0000 (03:44 +0000)]
code cleanup: some structs were declaring data when only typedef's were intended, make local vars and functions static.

5 years agocode cleanup: more zero as NULL pointers.
Campbell Barton [Wed, 7 Aug 2013 03:36:05 +0000 (03:36 +0000)]
code cleanup: more zero as NULL pointers.

5 years agocode cleanup: use NULL rather then zero for pointers
Campbell Barton [Wed, 7 Aug 2013 03:30:23 +0000 (03:30 +0000)]
code cleanup: use NULL rather then zero for pointers

5 years ago* Make MSVC happy.
Thomas Dinges [Tue, 6 Aug 2013 23:47:47 +0000 (23:47 +0000)]
* Make MSVC happy.

5 years agominor edits to float/double conversion suggested by DingTo
Campbell Barton [Tue, 6 Aug 2013 23:34:47 +0000 (23:34 +0000)]
minor edits to float/double conversion suggested by DingTo

5 years agoBGE Fix [#33215] KX_MeshProxy.transformUV() argument ValueError
Daniel Stokes [Tue, 6 Aug 2013 21:52:05 +0000 (21:52 +0000)]
BGE Fix [#33215] KX_MeshProxy.transformUV() argument ValueError

5 years agofix own regression in edgeloop delete [#36389] loop detete crashes
Campbell Barton [Tue, 6 Aug 2013 17:45:58 +0000 (17:45 +0000)]
fix own regression in edgeloop delete [#36389] loop detete crashes

5 years agoUsual edits/fixes to new UI messages...
Bastien Montagne [Tue, 6 Aug 2013 14:55:00 +0000 (14:55 +0000)]
Usual edits/fixes to new UI messages...

5 years agoFix for an link error in makesrna with SCons. Reported by Thomas Dinges, thanks!
Tamito Kajiyama [Tue, 6 Aug 2013 11:21:47 +0000 (11:21 +0000)]
Fix for an link error in makesrna with SCons.  Reported by Thomas Dinges, thanks!

Also removed some leftover of the Freestyle trunk merger in the makesrna subdirectory.

5 years agocode cleanup: remove lock from ViewerOperation class
Campbell Barton [Tue, 6 Aug 2013 09:27:20 +0000 (09:27 +0000)]
code cleanup: remove lock from ViewerOperation class

5 years agoview3d: gridfloor subdivisions wasn't respected for perspective mode.
Campbell Barton [Tue, 6 Aug 2013 07:10:18 +0000 (07:10 +0000)]
view3d: gridfloor subdivisions wasn't respected for perspective mode.

5 years agoquiet double-promotion warnings, change octree.cpp to use a float (vector accumulated...
Campbell Barton [Tue, 6 Aug 2013 06:38:52 +0000 (06:38 +0000)]
quiet double-promotion warnings, change octree.cpp to use a float (vector accumulated into a float anyway)

5 years agocycles builds with -Wdouble-promotion again.
Campbell Barton [Tue, 6 Aug 2013 06:36:34 +0000 (06:36 +0000)]
cycles builds with -Wdouble-promotion again.

5 years agoFix crash loading .blend files that have multiple outliners in a screen, saved
Brecht Van Lommel [Tue, 6 Aug 2013 05:35:54 +0000 (05:35 +0000)]
Fix crash loading .blend files that have multiple outliners in a screen, saved
with recent outliner optimizations (revision 58855, 3 days ago), in any Blender
version before this commit. On current Blender it would give a double free
warning in the console.

The problem is that it creates a temporary TreeStore on file save. However if you
have multiple outliners this memory block would always be at the same memory
address making it no longer unique. That then meant old memory address lookups on
file read were failing. Solution now is to postpone freeing these temporary memory
blocks until the end so that they are at unique addresses.

5 years agoselect faces after dissolving
Campbell Barton [Tue, 6 Aug 2013 05:05:31 +0000 (05:05 +0000)]
select faces after dissolving

5 years agoAdd key shortcut for context sensitive dissolve, Ctrl+X / Ctrl+Delete (as with...
Campbell Barton [Tue, 6 Aug 2013 05:01:11 +0000 (05:01 +0000)]
Add key shortcut for context sensitive dissolve, Ctrl+X  /  Ctrl+Delete  (as with node editor)

5 years agofix for using uninitialized boolean vars for mask select more/less
Campbell Barton [Tue, 6 Aug 2013 03:45:11 +0000 (03:45 +0000)]
fix for using uninitialized boolean vars for mask select more/less

5 years agocode clenup: remove benchmarking left in by accident and GPU print, also some minor...
Campbell Barton [Tue, 6 Aug 2013 02:47:47 +0000 (02:47 +0000)]
code clenup: remove benchmarking left in by accident and GPU print, also some minor style edits

5 years agoAdd assert to mul_v3_m3v3 and mul_v2_m3v3,
Sergey Sharybin [Tue, 6 Aug 2013 02:37:02 +0000 (02:37 +0000)]
Add assert to mul_v3_m3v3 and mul_v2_m3v3,
So they're not likely to be called with bad arguments.

5 years agoFrame reporting to console was wrong when not using Render Layers in compo
Sergey Sharybin [Tue, 6 Aug 2013 02:34:50 +0000 (02:34 +0000)]
Frame reporting to console was wrong when not using Render Layers in compo

This is because render info's current frame is either happens in
conversion stage (when using Render Layers node) or in sequencer's
rendering (which you don't usually have).

Now made it so when only compositor is used, proper current frame
is being set to re->i stats structure.

5 years agofix [#36363] ray_cast face index with n_gons
Campbell Barton [Tue, 6 Aug 2013 02:16:29 +0000 (02:16 +0000)]
fix [#36363] ray_cast face index with n_gons

5 years agoBGE: Making sure the BlenderPlayer calls Py_Finalize().
Mitchell Stokes [Tue, 6 Aug 2013 02:05:32 +0000 (02:05 +0000)]
BGE: Making sure the BlenderPlayer calls Py_Finalize().

5 years agoreplace mul_serie_m4 with mul_m4_m4m4 when only 3 args are given
Campbell Barton [Tue, 6 Aug 2013 01:45:29 +0000 (01:45 +0000)]
replace mul_serie_m4 with mul_m4_m4m4 when only 3 args are given

5 years agouse gcc malloc attribute for low level allocation functions, prevents gcc from checki...
Campbell Barton [Mon, 5 Aug 2013 20:57:13 +0000 (20:57 +0000)]
use gcc malloc attribute for low level allocation functions, prevents gcc from checking if resulting pointers alias existing pointers, also use sentinel attribute for uiButGetStrInfo so incorrect usage gives a warning.

5 years agoSplitViewer node:
Sv. Lockal [Mon, 5 Aug 2013 19:16:52 +0000 (19:16 +0000)]
SplitViewer node:
- fix thumbnail preview (previously it showed only one input)
- make SplitViewer node update even if the second input is not connected
- now it works when the first socket is connected to a zero-sized node tree (e. g. Color Input node)
- SplitViewer node is now based on 2 operations: SplitOperation and ViewerOperation.
- ViewerBaseOperation was removed as a redundant one. Any future viewer style node can use the same principle and prepare the output before passing to an actual ViewerOperation.

Thanks Lukas Toenne for reviewing this patch and giving me get few pieces of advice.

5 years agoAVI JPEG: remove the restriction to write only sizes that are multiples of 16.
Brecht Van Lommel [Mon, 5 Aug 2013 18:02:21 +0000 (18:02 +0000)]
AVI JPEG: remove the restriction to write only sizes that are multiples of 16.
Other encoders do not seem to have this restriction, and multiple video players
can play the files fine.

This also removes the same restriction for reading files, which actually caused
errors on some files with odd width/height.

5 years agoFix #36380: cycles render error with sun lamp that has object scale 0,0,0.
Brecht Van Lommel [Mon, 5 Aug 2013 14:22:07 +0000 (14:22 +0000)]
Fix #36380: cycles render error with sun lamp that has object scale 0,0,0.

5 years agoCycles OSL: image texture lookup optimization, acquire the per thread handle
Brecht Van Lommel [Mon, 5 Aug 2013 12:49:15 +0000 (12:49 +0000)]
Cycles OSL: image texture lookup optimization, acquire the per thread handle
for texture system in advance. Patch by Martijn Berger, with some tweaks.

There was about a 10% performance improvement on OS X in my tests with the
images.blend test file. This may be less on other platforms because OS X has
particularly slow mutex locks.

5 years agoFix build warnings with clang and UNUSED_RESULT_ATTR (attribute declaration must...
Brecht Van Lommel [Mon, 5 Aug 2013 12:49:13 +0000 (12:49 +0000)]
Fix build warnings with clang and UNUSED_RESULT_ATTR (attribute declaration must precede definition).

5 years agoFix wrong cycles cmake + msvc build flags with relwithdebinfo and minsizerel.
Brecht Van Lommel [Mon, 5 Aug 2013 12:02:43 +0000 (12:02 +0000)]
Fix wrong cycles cmake + msvc build flags with relwithdebinfo and minsizerel.
Patch by Karsten Schwenk.

5 years agofix for divide by zero when border render size causes zero pixel width
Campbell Barton [Mon, 5 Aug 2013 06:08:58 +0000 (06:08 +0000)]
fix for divide by zero when border render size causes zero pixel width

5 years agofix for blender-internal viewport render looking fuzzy, don't filter the image
Campbell Barton [Mon, 5 Aug 2013 05:13:09 +0000 (05:13 +0000)]
fix for blender-internal viewport render looking fuzzy, don't filter the image

5 years agoborder render with cycles had 1 pixel offset on the top-right edge of the image,...
Campbell Barton [Mon, 5 Aug 2013 04:52:27 +0000 (04:52 +0000)]
border render with cycles had 1 pixel offset on the top-right edge of the image, issue was caused by wmSubWindowScissorSet adding 1 to the ar->drawrct, now only add the padding when drawing the entire area

5 years agominor glitch with border drawing, use glRecti rather then glRectf, so rounding matche...
Campbell Barton [Mon, 5 Aug 2013 04:35:28 +0000 (04:35 +0000)]
minor glitch with border drawing, use glRecti rather then glRectf, so rounding matches the border render

5 years agouse api function for rect intersection
Campbell Barton [Mon, 5 Aug 2013 04:19:34 +0000 (04:19 +0000)]
use api function for rect intersection

5 years agoCode cleanup / Cycles:
Thomas Dinges [Mon, 5 Aug 2013 01:24:41 +0000 (01:24 +0000)]
Code cleanup / Cycles:
* Various string and comment fixes.

5 years agomempool api cleanup: differentiate mempool functions that allocate a pointer lookup...
Campbell Barton [Sun, 4 Aug 2013 19:40:50 +0000 (19:40 +0000)]
mempool api cleanup: differentiate mempool functions that allocate a pointer lookup table from allocating the data as a contiguous array,
call these functions BLI_mempool_as_table(), BLI_mempool_as_array(), the N prefixed versions of these functions return newly allocated arrays.

5 years agocode cleanup: remove redundant NULL checks
Campbell Barton [Sun, 4 Aug 2013 19:13:07 +0000 (19:13 +0000)]
code cleanup: remove redundant NULL checks

5 years agoadd c++/guardedalloc to more freestyle classes, also remove redundant 'public:'
Campbell Barton [Sun, 4 Aug 2013 18:50:00 +0000 (18:50 +0000)]
add c++/guardedalloc to more freestyle classes, also remove redundant 'public:'

5 years agocode cleanup: replace strcpy with BLI_strncpy for fixed size strings, and correct...
Campbell Barton [Sun, 4 Aug 2013 18:12:49 +0000 (18:12 +0000)]
code cleanup: replace strcpy with BLI_strncpy for fixed size strings, and correct some harmless but incorrect sizeof() use

5 years agofix for 2 memory leaks in dualcon library, quite bad since they leaked on every evalu...
Campbell Barton [Sun, 4 Aug 2013 18:05:29 +0000 (18:05 +0000)]
fix for 2 memory leaks in dualcon library, quite bad since they leaked on every evaluation.

5 years agoSupport WITH_CXX_GUARDEDALLOC for dualcon library
Campbell Barton [Sun, 4 Aug 2013 17:58:17 +0000 (17:58 +0000)]
Support WITH_CXX_GUARDEDALLOC for dualcon library

5 years agomore consistent use of checks of BLI_open(), check 'fd < 0' rather then -1. packedfil...
Campbell Barton [Sun, 4 Aug 2013 17:30:47 +0000 (17:30 +0000)]
more consistent use of checks of BLI_open(), check 'fd < 0' rather then -1. packedfile incorrectly treated 0 as an error value. best not be vague/sloppy with this.

5 years agofix memory leak in KX_ObstacleSimulation
Campbell Barton [Sun, 4 Aug 2013 17:20:03 +0000 (17:20 +0000)]
fix memory leak in KX_ObstacleSimulation

5 years agofix for BL_ArmatureConstraint::UpdateTarget(), was getting the pose from the target...
Campbell Barton [Sun, 4 Aug 2013 16:48:10 +0000 (16:48 +0000)]
fix for BL_ArmatureConstraint::UpdateTarget(), was getting the pose from the target, not the subtarget.

5 years agoCycles / OpenCL:
Thomas Dinges [Sun, 4 Aug 2013 12:40:46 +0000 (12:40 +0000)]
Cycles / OpenCL:
* Was broken after Blackbody merge, forgot to rename a struct...

5 years agoadd missing NULL check in RB_dworld_export if fopen fails.
Campbell Barton [Sun, 4 Aug 2013 04:30:14 +0000 (04:30 +0000)]
add missing NULL check in RB_dworld_export if fopen fails.

5 years agofix read outside buffer range KX_ObstacleSimulationTOI_rays::sampleRVO,
Campbell Barton [Sun, 4 Aug 2013 04:07:29 +0000 (04:07 +0000)]
fix read outside buffer range KX_ObstacleSimulationTOI_rays::sampleRVO,
Was using 3d vectors for 2d operations, passing float[2] to args that use MT_Vector3 was reading the 3rd value of a 2d array

5 years agoKX_ObstacleSimulation: replace inline math functions with BLI_math functions
Campbell Barton [Sun, 4 Aug 2013 03:47:43 +0000 (03:47 +0000)]
KX_ObstacleSimulation: replace inline math functions with BLI_math functions

5 years agocode cleanup: replace strncpy -> BLI_strncpy, skip first 2 chars when making a hash...
Campbell Barton [Sun, 4 Aug 2013 03:45:30 +0000 (03:45 +0000)]
code cleanup: replace strncpy -> BLI_strncpy, skip first 2 chars when making a hash from the object name.

5 years agofix navmesh, error was introduced by r58420 looks like changes weren't tested.
Campbell Barton [Sun, 4 Aug 2013 03:43:02 +0000 (03:43 +0000)]
fix navmesh, error was introduced by r58420 looks like changes weren't tested.

5 years agoavoid freeing NULL pointer for navmesh BuildVertIndArrays and dont truncate memset...
Campbell Barton [Sun, 4 Aug 2013 03:02:35 +0000 (03:02 +0000)]
avoid freeing NULL pointer for navmesh BuildVertIndArrays and dont truncate memset arg.

5 years agoavoid using MEM_reallocN_id directly, add utility macro for freeing.
Campbell Barton [Sun, 4 Aug 2013 03:00:04 +0000 (03:00 +0000)]
avoid using MEM_reallocN_id directly, add utility macro for freeing.

5 years agoavoid runtime overflow (1 << 31), for RNA and armature layer UI.
Campbell Barton [Sun, 4 Aug 2013 00:01:41 +0000 (00:01 +0000)]
avoid runtime overflow (1 << 31), for RNA and armature layer UI.

5 years agocode cleanup: remove unused define, correct header guard mismatch, add NULL check...
Campbell Barton [Sat, 3 Aug 2013 23:58:17 +0000 (23:58 +0000)]
code cleanup: remove unused define, correct header guard mismatch, add NULL check so DM_to_mesh() can have a NULL object passed (currently not used) and remove redundant NULL check.

5 years agofix for strange error with BGE 2d filter code,
Campbell Barton [Sat, 3 Aug 2013 23:40:15 +0000 (23:40 +0000)]
fix for strange error with BGE 2d filter code,
RAS_2DFilterManager::RenderFilters was casting an int to an unsigned int, then doing a subtraction which would give a negative number.

5 years agominor changes to BLI_heap, save some CPU cycles.
Campbell Barton [Sat, 3 Aug 2013 22:04:47 +0000 (22:04 +0000)]
minor changes to BLI_heap, save some CPU cycles.
added an assert for incorrect use of BLI_heap_remove

5 years agoadd missing NULL checks from BKE_constraint_get_typeinfo(), so constraints from the...
Campbell Barton [Sat, 3 Aug 2013 22:03:15 +0000 (22:03 +0000)]
add missing NULL checks from BKE_constraint_get_typeinfo(), so constraints from the future dont crash.
also remove some redundant NULL checks.

5 years agoCycles / SSS passes:
Thomas Dinges [Sat, 3 Aug 2013 21:56:27 +0000 (21:56 +0000)]
Cycles / SSS passes:
* Connecting the new passes in the compositor caused a crash, forgot to register the new passes in the compositor operations code.

5 years agoCycles:
Thomas Dinges [Sat, 3 Aug 2013 21:45:57 +0000 (21:45 +0000)]
Cycles:
* Forgot to rename some SSS pass strings.
* Some typo fixes.

5 years agobmesh: improve limited dissolve result
Campbell Barton [Sat, 3 Aug 2013 21:01:42 +0000 (21:01 +0000)]
bmesh: improve limited dissolve result

iteratively dissolve the best edge/vert, updating the heap as the dissolve runs.

5 years agofix for over-allocation in BKE_pbvh_search_gather, BKE_pbvh_gather_proxies,
Campbell Barton [Sat, 3 Aug 2013 18:05:30 +0000 (18:05 +0000)]
fix for over-allocation in BKE_pbvh_search_gather, BKE_pbvh_gather_proxies,
each element was having the size of PBVHNode allocated rather then the size of a pointer (8 vs 184 bytes here)

5 years agoadd versions of MEM_reallocN, MEM_recallocN which take a string arg so new allocs...
Campbell Barton [Sat, 3 Aug 2013 17:53:41 +0000 (17:53 +0000)]
add versions of MEM_reallocN, MEM_recallocN which take a string arg so new allocs have an ID, changing existing functions signatures would be too disruptive at the moment.

5 years agocorrect edgeloop tagging assigning an uninitialized value, also remove redundant...
Campbell Barton [Sat, 3 Aug 2013 17:29:53 +0000 (17:29 +0000)]
correct edgeloop tagging assigning an uninitialized value, also remove redundant NULL check in edgenet fill

5 years agocorrect invalid sizeof() use in bmesh (harmless in practice)
Campbell Barton [Sat, 3 Aug 2013 17:27:05 +0000 (17:27 +0000)]
correct invalid sizeof() use in bmesh (harmless in practice)

5 years agofix error in pose bone selection - head/tail mixup
Campbell Barton [Sat, 3 Aug 2013 17:10:00 +0000 (17:10 +0000)]
fix error in pose bone selection - head/tail mixup

5 years agocode cleanup: remove duplicate checks
Campbell Barton [Sat, 3 Aug 2013 16:55:49 +0000 (16:55 +0000)]
code cleanup: remove duplicate checks

5 years agocode cleanup: bmesh use 'const' for query functions.
Campbell Barton [Sat, 3 Aug 2013 16:37:23 +0000 (16:37 +0000)]
code cleanup: bmesh use 'const' for query functions.

5 years agocode cleanup: replace bmesh_radial_face_find -> BM_edge_in_face
Campbell Barton [Sat, 3 Aug 2013 15:30:57 +0000 (15:30 +0000)]
code cleanup: replace bmesh_radial_face_find -> BM_edge_in_face

5 years agoFix #36288, Renderlayer toggles behave unexpected in Outliner. The outliner was using...
Lukas Toenne [Sat, 3 Aug 2013 15:00:22 +0000 (15:00 +0000)]
Fix #36288, Renderlayer toggles behave unexpected in Outliner. The outliner was using the wrong flag variable for toggling render layers, was setting pass flags instead of the layer on/off flag.

5 years agoMerged revision(s) 58857 from trunk/blender into soc-2013-dingto.
Thomas Dinges [Sat, 3 Aug 2013 13:13:53 +0000 (13:13 +0000)]
Merged revision(s) 58857 from trunk/blender into soc-2013-dingto.

Note: For next merge skip 58858, SSS passes merge to trunk.

5 years agoCycles / SSS:
Thomas Dinges [Sat, 3 Aug 2013 13:12:09 +0000 (13:12 +0000)]
Cycles / SSS:
* Render Passes are now available for Subsurface Scattering (Direct, Indirect and Color pass).

This is part of my GSoC project, SVN merge of r58587, r58828 and r58835.

5 years agoskip BLI_STATIC_ASSERT for Coverity builds (caused parse error)
Campbell Barton [Sat, 3 Aug 2013 13:08:51 +0000 (13:08 +0000)]
skip BLI_STATIC_ASSERT for Coverity builds (caused parse error)

5 years agoMerged revision(s) 58785-58855 from trunk/blender into soc-2013-dingto.
Thomas Dinges [Sat, 3 Aug 2013 12:49:08 +0000 (12:49 +0000)]
Merged revision(s) 58785-58855 from trunk/blender into soc-2013-dingto.

5 years agofix for [#36260] 2,300 Objects Makes Blender Unresponsive
Sv. Lockal [Sat, 3 Aug 2013 11:35:09 +0000 (11:35 +0000)]
fix for [#36260] 2,300 Objects Makes Blender Unresponsive

- performance of outliner was low because of unoptimal data structures.
- now it uses BLI_mempool instead of custom mempool and GHash to make searches for duplicates faster.
- also fix undesired behaviour of BLI_mempool_as_arrayN

thanks to Campbell Barton and Lukas Tönne for helping me get a better fix put together.

5 years agoDynamic Paint: Added a new "smoothness" parameter for waves.
Miika Hamalainen [Sat, 3 Aug 2013 09:46:38 +0000 (09:46 +0000)]
Dynamic Paint: Added a new "smoothness" parameter for waves.

It greatly helps getting rid of that "noise" that occurs if you use really steep objects (like cubes) as a brush. New default value is 1.0 which is just high enough to only get rid of the sharpest spikes, so if you want really smooth waves it's better use higher values.

This also seems to "fix" bug [#35413].

5 years agoBGE: Making sure m_drawingmode is initialized in the various RAS_Storage constructors.
Mitchell Stokes [Sat, 3 Aug 2013 05:02:03 +0000 (05:02 +0000)]
BGE: Making sure m_drawingmode is initialized in the various RAS_Storage constructors.

5 years agoBGE: Making sure m_left, m_right, m_parent, m_radius, and m_client_object are initial...
Mitchell Stokes [Sat, 3 Aug 2013 05:01:57 +0000 (05:01 +0000)]
BGE: Making sure m_left, m_right, m_parent, m_radius, and m_client_object are initialized in the SG_Tree constructors.

5 years agoBGE: Making sure m_objType is initialized in the BlendType (VideoTexture) constructor.
Mitchell Stokes [Sat, 3 Aug 2013 05:01:50 +0000 (05:01 +0000)]
BGE: Making sure m_objType is initialized in the BlendType (VideoTexture) constructor.

5 years agoBGE: Making sure m_line is initialized in the Exception (VideoTexture) constructor.
Mitchell Stokes [Sat, 3 Aug 2013 05:01:45 +0000 (05:01 +0000)]
BGE: Making sure m_line is initialized in the Exception (VideoTexture) constructor.

5 years agoBGE: Making sure m_buffV, m_buffU, and m_pitchUV are initialized in the FilterYV12...
Mitchell Stokes [Sat, 3 Aug 2013 05:01:40 +0000 (05:01 +0000)]
BGE: Making sure m_buffV, m_buffU, and m_pitchUV are initialized in the FilterYV12 constructor.

5 years agoBGE: Making sure m_offset is initialized in the ImageSourceMix constructor.
Mitchell Stokes [Sat, 3 Aug 2013 05:01:36 +0000 (05:01 +0000)]
BGE: Making sure m_offset is initialized in the ImageSourceMix constructor.

5 years agoBGE: Making sure m_mirrorHalfWidth and m_mirrorHalfHeight are initialized in the...
Mitchell Stokes [Sat, 3 Aug 2013 05:01:32 +0000 (05:01 +0000)]
BGE: Making sure m_mirrorHalfWidth and m_mirrorHalfHeight are initialized in the ImageRender constructor.

5 years agoBGE: Removing some unreachable and unused code in BL_ActionManager.cpp.
Mitchell Stokes [Sat, 3 Aug 2013 05:01:23 +0000 (05:01 +0000)]
BGE: Removing some unreachable and unused code in BL_ActionManager.cpp.

5 years agoBGE: Fixing a memory leak with PyType_Ready_ADD.
Mitchell Stokes [Fri, 2 Aug 2013 22:23:42 +0000 (22:23 +0000)]
BGE: Fixing a memory leak with PyType_Ready_ADD.

5 years agoCycles / SSS Render Passes:
Thomas Dinges [Fri, 2 Aug 2013 19:57:14 +0000 (19:57 +0000)]
Cycles / SSS Render Passes:
* Finished the implementation, Direct and Indirect Passes work now.

Patch by Brecht and myself.

5 years agoFix [#35541]: Point cache doesn't load existing frames from disk if cache file name...
Miika Hamalainen [Fri, 2 Aug 2013 16:39:50 +0000 (16:39 +0000)]
Fix [#35541]: Point cache doesn't load existing frames from disk if cache file name is changed.

5 years agofix(Collada): wrong usage of the set attribute with multiple UV sets
Gaia Clary [Fri, 2 Aug 2013 15:58:11 +0000 (15:58 +0000)]
fix(Collada): wrong usage of the set attribute with multiple UV sets

5 years agoCycles / Subsurface Render Passes:
Thomas Dinges [Fri, 2 Aug 2013 14:40:21 +0000 (14:40 +0000)]
Cycles / Subsurface Render Passes:
* Code cleanup, "subs" abbreviation is not really meaningful, so let's write "subsurface" instead.

5 years agocorrect error in own recent edits to triangle joining
Campbell Barton [Fri, 2 Aug 2013 13:35:04 +0000 (13:35 +0000)]
correct error in own recent edits to triangle joining

5 years agocode cleanup:
Campbell Barton [Fri, 2 Aug 2013 13:21:32 +0000 (13:21 +0000)]
code cleanup:
- incorrect NULL check in logic UI drawing
- incorrect NULL check in octree quad test

5 years agoSome versioning changes/patching of the brush system:
Antony Riakiotakis [Fri, 2 Aug 2013 00:24:34 +0000 (00:24 +0000)]
Some versioning changes/patching of the brush system:

* Change overlay alphas to 33 if not initialized. This should have been
done for 2.67 but better do it now to avoid frustration with overly
transparent overlays. For users that have set this low manually this
will reset the setting to 33 but I think it's less irritating than users
who unwillingly had the setting to 1 and saw nothing when activating the
overlay.

* Allow overlay alpha to be zero as well

* Reset old, now obsolete BRUSH_FIXED flag for brushes that still use
this. I am doing this here to avoid patching hell when the paint brush
is merged, since it's not possible to know the merged version
beforehand, and this flag will be used.