Implemented basic audio animation.
* AnimatableProperty: Propper cache writing and spline interpolation for reading (the solution for stair steps in audio animation)
* Animatable properties so far are: volume, pitch, panning
* Users note: Changing the pitch of a sound results in wrong seeking, due to the resulting playback length difference.
* Users note: Panning only works for mono sources, values are in the range [-2..2], this basically controls the angle of the sound, 0 is front, -1 left, 1 right and 2 and -2 are back. Typical stereo panning only supports [-1..1].
* Disabled animation of audio related ffmpeg output parameters.
* Scene Audio Panel: 3D Listener settings also for Renderer, new Volume property (animatable!), Update/Bake buttons for animation problems, moved sampling rate and channel count here
Main: Complete rewrite of the sequencer related audio code to support 3D Audio objects later and especially adressing the animation system problems (see mailing list if interested).
Note: Animation is not working yet, so with this commit volume animation doesn't work anymore, that's the next step.
Minor things:
* Changed AUD_Reference behaviour a little to be more usage safe.
* Fixed bug in AUD_I3DHandle: Missing virtual destructor
* Fixed enmus in AUD_Space.h
* Fixed a warning in rna_scene.c
* Removed an unneeded call in rna_sound.c
- Default size of Graph Editor handle vertices is now 4 (up from 3).
This "small" change seems to be enough to make a substantial
difference when it comes to actually differentiating between these
- "Only Selected" DopeSheet filter is enabled for new Graph Editor
instances by default. It helps hone in on the F-Curves of the data
most animators hope to just be refining the motion for (i.e. the
selected stuff)
- "Only Selected Keyframe Handles" is now enabled, to reduce clutter
from handles of nearby keys getting in the way.
Bezier curves are now drawn smoother (i.e. less segmented), especially
for curve segments where there is a very large vertical displacement
over a short period of time (i.e. 120 degrees rotation over 1 frame)
and/or often when zoomed in a bit too.
- Made the resolution calculation take the vertical distance into
account too, instead of just the horizontal distance.
- Segment multiplier changed from 3 to 5, as this seems to give better
zoomed-in performance.
This fixes bug #26764 and several others like it, where modifier
properties (and others, but most visibly modifiers) would not do
anything when animated or driven, as modifier properties require the
RNA update calls to tag the modifiers to get recalculated.
While just adding a call to RNA_property_update() could have gotten
this working (as per the Campbell's patch attached in the report, and
also my own attempt #25881). However, on production rigs, the
performance cost of this is untenatable (on my own tests, without
these updates, I was getting ~5fps on such a rig, but only 0.9fps or
possibly even worse with the updates added).
Hence, this commit adds a property-update caching system to the RNA
level, which aims to reduce to the number of times that the update
functions end up needing to get called.
While this is much faster than without the caching, I also added an
optimisation for pose bones (which are numerous in production rigs) so
that their property updates are skipped, since they are useless to the
animsys (they only tag the depsgraph for updating). This gets things
moving at a more acceptable framerate.
* Reviewed and improved the linear resampler. Now it should work pretty good also for special cases that caused errors previously.
* Fixed a crash in the GE when a sound actuator doesn't have a sound assigned.
* Corrected the OpenAL device's threading code. This is a bugfix for #27913, thanks to Juha Mäki-Kanto for helping to resolve this.
- Intersection code was using undefined vector
caused wrong lines to be picked
- Code now also copes with hidden sockets.
If all fails, is just unhides a good socket.
On dragging a non-connected node on a noodle, it will insert it.
Functionality tweaks are possible, but it already feels non-intrusive.
Rules:
- Insertion only when a single noodle is intersecting with node.
- Default connects first matching socket type.
- If no socket match, it connects the first.
Recoded side-chain reparenting step to fix (as far as I've been able
to tell) infinite looping problems which were a bit intermittent here
using the test file. The fix here involves some tighter checks to
prevent corrupting the parenting of bones in the run of bones being
merged but also of any ancestors of those.
Paste pose no longer just does a blind "replace all properties" on
bones that it pastes on. Instead:
* when properties exist on the target already - only change the
properties in common
* when properties don't already exist - copy all properties
Time-scale drawing wasn't respecting the time unit setting.
While working on this, I tried to tweak the grid drawing to a more
common setting. It's hardcoded to show lines at every 25 px = once
every 25 frames, which is only really fine when FPS=25. Anyways, this
works fine enough for the sequencer for now in general usage.
functions based on editor types
This could be split up further in future if there's such a need, but
this should already be sufficient. Most notably required since the NLA
recalc stuff was taking quite a few lines within that block
Blender render optimizes alpha=0 materials away, unless it has
a number of properties... but there wasn't a check for material
being ray-mirror, it then should be rendered always.
[#28032] Python Mathutils: Matrix Multiplication Error
Since they ended up reversing the order we better keep old code unless its proven to be incorrect.
also change Matrix.__repr__ function args to evaluate correctly (need to be inside a tuple).
From what I can tell there is no good fix for this bug, calculating the 2d/3d viewborder and then attempting to align them to be pixel perfect fails because of float imprecision.
Added a workaround, so the camera border is always drawn in 2d space,
since this workaround may cause problems later on its kept under the define VIEW3D_CAMERA_BORDER_HACK so we can get old behavior back easily.
Generated images would not be re-generated with a float buffer on load, even when selected on creation.
Now save the float buffer setting as a generated image flag.
This means you can enable before baking to enable baking to a float buffer.
rendering, to prevent any race condition problems
I've noticed some weird and random crashes recently while rendering,
which I suspect have been arising from having an Action Editor open
while rendering. Previously only the timeline was patched against
these problems, though the issues may be more widespread. Hence,
solving this problem at the root cause instead.
