Blender Git Loki

Cleanup: Don't use C++ comments style in C code

Fix for T50373: lattices should not be able to get subsurf modifiers

Reviewers: mont29

Reviewed By: mont29

Subscribers: sergey

Differential Revision: https://developer.blender.org/D2449

Fix T50385: Deadlock in BKE_libblock_remap_locked.

Am pretty sure node update should not touch to Main database like that,
but for now let's allow it, I guess the hack is needed for things like
Sverchok. ;)

Fix: Audio plays back incorrectly after rendering to a video file

D2365

Make button span the full width of the properties editor

Remove instances of old vector icons

I checked and this makes consistent with other area of blender.
See https://developer.blender.org/rBf86eccb1ca2388ae203df2a59285a2f2fb9b6aa8#131061

Merge branch 'master' into blender2.8

Conflicts:
source/blender/editors/gpencil/drawgpencil.c

Show 'Manipulate Center Points' in weight paint mode

If the active object is in weight paint mode, but some armatures in pose mode, 'manipulate center points' still affects the transformation. See bd2034a749a9a9.

Also removed redundant check, we basically did the same check for paint modes twice.

Dynamic Paint: Fix random pixel flooding by absolute brush with spread.

If a very low wetness absolute alpha brush is used with spread and
drying effects enabled, some pixels will rapidly accumulate paint.
This happens because paint drying code applies a minimal wetness
threshold that causes the paint to instantly dry out.

Specifically, every frame the brush adds paint at the specified
absolute alpha and wetness set to the minimal threshold, spread
drops it below threshold, and finally drying moves all paint to
the dry layer. This drastically accelerates the rate of flow of
paint into the affected pixels.

Fortunately, the reason paint spread actually ends up decreasing
wetness turns out to be a simple floating point precision problem,
which can be easily fixed by restructuring the affected expression.

Fix (IRC reported) bad handling of Text data-block user count.

Reported on IRC by dfelinto, thanks.

Root of the issue was that opening a new text file would create
datablock with one user, when Text editor is actually a 'user one' user.

This was leaving Text datablocks in inconsitent user count, and
generating asserts in BKE_library area.

Also changed a weird piece of code related to that extra user thing in
main remapping func.

Fix T50369: Objects can't be deleted from scene when using "link group objects to scene"

Main issue here was that in old usercount system 'user_real' did simply
not allow that kind of thing to work. With new pait of 'USER_EXTRA'
tags, it becomes possible to handle the case correctly, by merely refining
checks about indirectly use objects whene removing them from a scene.

Incidently, found another related bug, 'link group objects to scene' was not
incrementing objects' usercount - bad, very very bad!

Fix T50350: Quick Explode time frame problem

The settings.frame_start rna was clamping frame start to frame end when frame start was bigger than frame end.
The fix is simply to set frame end first

Revert "Remove double menu entries"

This reverts commit e2d7efc9503720770b9652284e25006eecbec8d4.

Fix implicit size_t to int conversion.

Seems like it was erroring on some buildbots...

Fix: NLA "Strip Time" setting cannot be edited

This is a hacky fix for a regression introduced sometime after 2.76.
The "Strip Time" setting on NLA Strips could not be edited without the
value immediately jumping back to the current FCurve value (or 0.0 if no
keyframes existed); even enabling autokey wouldn't let you key the property.

Until we have proper overrides (that only lose their values on frame change),
it's best that this setting is editable, even if it does mean it you have to
manually change the frame to see the updated values.

Merge remote-tracking branch 'origin/master' into blender2.8

I'm deliberately ignoring/reverting rB070f22c4 which in turns means T49861 is happening on 2.8

GPencil: Per-layer option to always show onion skinning

Sometimes it can be useful to be able to keep onion skins visible in the
OpenGL renders and/or when doing animation playback. In particular, there
are two use cases where this is quite useful:
1) For creating a cheap motion-blur effect, especially when the before/after
values are also animated.
2) If you've animated a shot with onion skinning enabled, the poses may end
up looking odd if the ghosts are not shown (as you may have been accounting
for the ghosts when making the compositions).

This option can be found as the small "camera" toggle between the "Use Onion Skinning"
and "Use Custom Colors" options.

Fix: Passed wrong last argument to function

Merge remote-tracking branch 'origin/master' into blender2.8

Fix T49861: Interlace stereo is broken in 2.78

This is a regression introduced in rB5bd9e832

It looks more like a hack than a proper fix, but the shader logic
changed a lot for blender2.8, so I would rather do the elegant fix
there, while leaving master working.

If we ever do a 2.78b (or 2.79) this should get in.