This fixes again some segfaults on error conditions.
It also gives some reasonable error-messageboxes if chooses the wrong
DV-format-options.
More to come...
Bugfixes:
- hddaudio now allocates it's decode buffer + AVCODEC_MAX_AUDIO_FRAME_SIZE
safety interval. (The former code expected all decoded audio frames to
be the same size which can hurt under some circumstances e.g. VBR files)
- writeffmpeg: some pointers where not initialized on start and after
deletion not set to null. Could segfault randomly on error conditions.
- drawseq: use startdisp and enddisp to decide, if a strip should be drawn.
(Just extend strips first or last frame beyond screen dimensions in
the previous version and watch the strip disappear)
Small fix, but results are at least less frustrating now. It uses the
"compatible euler" function from inserting key positions here, preventing
euler values to be constrainted that differ weirdly.
I've tried several other approaches to get a definite rotate constraint,
but only constraining a single axes seems to me impossible magic still...
The new (months ago) added option for "Local Constraint" always copied
and restored a PoseChannel quaternion, that disabled the required
normalizing of quaternions.
Only visible when you manually draw Ipo Curves, but it's probable that
it showed in other cases too.
Object Layer Ipos didn't work when the Ipo was moved an Object Action.
Can't get this to work though... this option has been added with a lot
of exception handling already, and for action/nla it's very nasty to
blend/mix/add layer bit values. So; better not support that.
As feedback for users, I've added a notice popup when you try to move
layer-ipos to an action.
Compositor:
Viewer Nodes inside of Groups now work too. To not frustrate interactive
speed, the following rules apply:
- Making a Group editable (or closing) doesn't signal recalculation of
composite
- clicking on an Input socket always checks changes and calulcates
- When there are Viewers inside an edited Group, Viewers in the main tree
are not executed.
Also added: a "hide unused sockets" icon in the header of Viewer nodes.
This allows cleanup of Groups, to prevent these sockets get reveiled.
Each report was about a different failure with Particles, all related to
weak handling of animation systems and the depsgraph.
Fix has 2 parts; depsgraph now signals "object recalc" to be for time
changes; this then is used to bypass particle-building (since that's baked).
Other part is better object caching while makig particles.
- Shaded drawmode is back (shift+z).
Note it still only uses orco texture; but lighting/shading is using
the internal render module entirely.
- "Make Sticky" option back.
(Also fix in sticky texture render, was wrong scaled)
This commit brings back:
- Field Render
- MBlur Render (old style)
- Border render with or without cropping
Note: Field Render is not supported in Compositor yet. Blurring or filter
will destroy field information.
Both MotionBlur as Field render are done before Compositing happens.
Fixes:
- The "Save Buffers" option only worked on single frame renders, not for
Anim render.
- Found an un-initalized variable in Render initialize... this might have
caused the unknown random crashes with render.
Code restructure:
Cleaned up names and calls throughout the pipeline, more clearly telling
what goes on in functions.
This is visible in the updated first image of the Wiki doc:
http://mediawiki.blender.org/index.php/BlenderDev/RenderPipeline
Material Nodes: The Texture node didn't do the standard "2d mapping" yet
in case an Image Texture is used. Caused wrong mapping for example for UV
coordinate inputs.
Groups are supposed to play its entire (internal) animation system
properly, also when using a dupli-group, linking the group from other file.
When dependencies in a scene change, the groups didn't get corrected,
causing dupli-group to not playback without lag.
This commit resorts the group objects on each DAG_scene_sort() call.
- code for detecting loops was overly complex, and detected loops wrong
even (when chains partially overlap it gave error)
- that code could NULLify sets of other scenes than the one you tested
- and that code should be in blenkernel, so it can be called in more
locations, like fixing potential loops in old files
- the code for looping over the chained scenes was bad too, I created a
nice #define to keep code readable. (See SETLOOPER in BKE_scene.h)
ALso fixes reports in tracker :)
CVS these days in order not to be terribly annoyed by its absence, here it
is: instinctive-blender's "Recent files list".
It's in the CTRL-O menu. No UI / muscle memory changes -- the first entry is
the same as the only entry that used to be in the former popup.
.
Added simplified fix as proposed by Stephan Kassemeyer to allow scaled
Armatures to still have a working 'stride bone'. Only works for uniform
scaled armatures though.
In general, for properly working armatures, I'd recommend to never scale
it at all, and certainly not non-uniform scaling. It will give issues with
constraints, IK and drivers all over...
This will free all Node output buffers, while compositing, when not needed
anymore. Saves a whole lotta memory, and will enable to use many many more
nodes (or high resolution images).
- Found several places, where people explicitly casted the frame number
to short.
- Fixed the crash in BPY_interface by adding an empty line (to make it
recompile everywhere, make clean doesn't help...)
For the build system maintainers:
Problem was: The change in makesdna changed the position of the
scriptlink structure. BPY_interface.c somehow didn't get recompiled
(not even after a make clean!!!) which triggered crashes on adding
scriptlinks.
previously would only work if the armature was the first in the meshes modifier list,
in that case the armature would be name flipped but the mesh would not)
When multiple output nodes exist (Material), the active Output flag could
get copied and wasn't reset properly. Now the depenendency sorting code
ensures only 1 output node is the active output for execution.
exit_usiblender() to finalize Python before main library data was freed.
This solved a somewhat specific sigsegv with pydrivers, but as Ken
Hughes found out (thanks!) caused one with scripts that called Blender.Exit().
Now running scripts (G.main->script) are freed in BPY_end_python()
itself (so before the rest of the library data is freed), before
Py_Finalize(). Works fine in all my tests so far.
The file script.c should become obsolete with this change (I added a
comment about it there). If all is indeed fine, it will be removed
later.
text module when user edits the input text box of any pydriver
(Transform Properties panel, Ipo window).
It's enough to click in and out of a single pydriver's text input box
for the module reloading and also re-evaluation of all pydrivers
available. Maybe this "refreshing" should also be available from a
menu, let's see.
Note for Python fans:
Definitions and redefinitions in a reloaded module are properly handled
in Python, but previously defined data in the module doesn't disappear.
So if you define a function "f" inside a module, import it, then change
the function's name to "g" and reload the module, both "f" and "g" will
be available. This is considered a feature, check reload's documentation:
http://docs.python.org/lib/built-in-funcs.html#l2h-59
wiki with info: http://mediawiki.blender.org/index.php/BlenderDev/PyDrivers
(there are two sample .blends in the patch tracker entry, last link in
the wiki page)
Notes:
In usiblender.c I just made Python exit before the main library gets
freed. I found a situation with pydrivers where py's gc tried to del
objects on exit and their ID's were not valid anymore (so sigsegv).
Ton needs to check the depsgraph part.
For now pydrivers can reference their own object, something normal
ipodrivers can't. This seems to work fine and is quite useful, but if
tests prove the restriction is necessary, we just need to uncomment a
piece of code in EXPP_interface.c, marked with "XXX".
Thanks Ton for the ipodrivers code and adding the hooks for the py part
and Martin for the "Button Python Evaluation" patch from which I started
this one.
Anyone interested, please check the wiki, the .blends (they have
README's) and tell me about any issue.
commented out some code that makes Blender crashing, after leaving the game engine (armature deletes some pose, which deletes constraints, which are still in the dependency graph. Ask Charlie)
When the 'reference shape key' (drawn yellow) was not the first key, the
channels as drawn in IpoWindow didn't match the actual shape keys.
This was caused by an exception in code that skips drawing the reference
shape when 'relative' was used.
Now I've added a rule that the first shape in a list always becomes the
reference, that way you can also edit it. To keep backwards compatibility,
this is only activated on translating the shape key lines.
Patch #4044 patch to change the step when changing frame with up or down arrow keys.
This adds a per Scene setting to specify the number of frames skipped by the up and down arrow.
This setting can be changed in the Timeline header and in the Anim/Playback section of the Scene settings.
Upon loading a file without that setting or creating a new scene, it is set to 10
Also fixed the Start and End frame buttons in the Scene buttons to do a proper refresh.
Using stride-bone in an NLA, on a path without speed Ipo, didn't correct
the case when an action starts on a non-zero value.
Patch provided by Roland Hess. Thanks!
The issue was that particle emittors were still transformed by the object
matrix itself. That was solved in the previous commit, but there was
still an error in correctly evaluating dependencies for the object...
Current commit uses depsgraph to recalculate all objects that influence
the emittor.
The depsgraph code doesn't like particles much (because it uses baking).
Current construct is still weak, is on the list to solve nice.
