Blender Git Commit Log

All Blender Git commits.

Page: 4276 / 8462

Revision 9ff8d4a by Mike Erwin (master)
July 21, 2016, 22:26 (GMT)
fix enum type / values mismatch

Found with PVS-Studio T48917
Revision 7c99b71 by Mike Erwin (master)
July 21, 2016, 22:16 (GMT)
fix null pointer dereferences

Found with PVS-Studio T48917
Revision 23f1b20 by Mike Erwin (master)
July 21, 2016, 22:16 (GMT)
fix comparison of identicals

Some of these check that dimensions match before running code that
assumes they do match.

For imb_stereo3d_write_anaglyph I *assume* this change reflects the
intended behavior. Before it was always grabbing alpha from buffer 0.

Found with PVS-Studio T48917
July 21, 2016, 21:43 (GMT)
Curves: Trim operator

The trim operator works fine in 1/3 cases.
July 21, 2016, 21:33 (GMT)
Cleanup: pass pointer to texture draw state
July 21, 2016, 21:22 (GMT)
BMesh: compare face angles as angles cosine
July 21, 2016, 21:11 (GMT)
Add an undef for sc_next for safety
July 21, 2016, 21:07 (GMT)
Fix broken proxy object deletion.

Check to ensure we do not delete last instancing of an indirectly used object (forbidden
because it creates 'ghost' objects user have no real way to re-instance) was defeated by
the backward pointer 'proxy_from', which generated a 'false' indirect usage of (local)
proxy object itself (the one we are trying to delete).

Fixed by actually considering that ID usage as local if proxy object itself is local
(because that pointer actually does not 'pertain' to the object holding it, but to its proxy).

Yeah, it's... complicated, twisted and ugly - it's proxy.
Revision 6ebce7e by Mike Erwin (master)
July 21, 2016, 19:53 (GMT)
fix typo in condition

(A - A).norm() is always 0 so condition is always true.

(A - B).norm() and (B - A).norm() both compute the same distance so I
picked one to match surrounding code.

Found with PVS-Studio T48917
July 21, 2016, 19:36 (GMT)
Fix T48915: Wrong threshold on scaled objects and ortho view.

`ray_is_normalized` == true in DerivedMesh
July 21, 2016, 18:22 (GMT)
Fix crash w/ auto-insert offset
July 21, 2016, 18:16 (GMT)
Each node gets expanded to two function calls for autodiff.

The value is calculated without any derivative inputs (this is not a differential equation).
The derivatives in both variables also depend on the partial derivatives of the inputs.
July 21, 2016, 18:06 (GMT)
Cleanup: warnings
Revision 9279bee by Mike Erwin (master)
July 21, 2016, 17:50 (GMT)
remove repeated codec-supports-alpha check

I don?t think any other codec enum makes sense here, so probably just
an extra copy/paste. Here are the video codecs:

AV_CODEC_ID_NONE
AV_CODEC_ID_MPEG4
AV_CODEC_ID_MJPEG
AV_CODEC_ID_DNXHD
AV_CODEC_ID_MPEG2VIDEO
AV_CODEC_ID_MPEG1VIDEO
AV_CODEC_ID_DVVIDEO
AV_CODEC_ID_THEORA
AV_CODEC_ID_PNG <? alpha
AV_CODEC_ID_QTRLE <? alpha
AV_CODEC_ID_FFV1 <? alpha (if enabled)
AV_CODEC_ID_HUFFYUV <? alpha
AV_CODEC_ID_H264
AV_CODEC_ID_FLV1

Found with PVS-Studio T48917
July 21, 2016, 16:32 (GMT)
WIP Simulated Annealing

Iterations now only recompute placement of one random chart according to the main reference paper
Also includes stubs for margin computation
July 21, 2016, 16:32 (GMT)
Incorporate the generated GLSL code for hair displacement into the strands vertex shader.
July 21, 2016, 16:13 (GMT)
Make sure edges are computed.
July 21, 2016, 16:03 (GMT)
Emit function calls for each node.

Base functions are yet to be defined.
July 21, 2016, 15:38 (GMT)
Support dual values for autodiff'ing in GLSL codegen right from the start.

Note that for the hair displacement function we would only be interested
in a single derivative (the tangent). For now we just use the existing
2-variable duals, later on this can be generalized to avoid unnecessary
and confusing code emission.
July 21, 2016, 15:27 (GMT)
Make mesh validation optional.
By: Miika HämäläinenLast update: Nov-07-2014 14:18MiikaHweb | 2003-2021