offsets too (like for location)
This is useful in some cases when Copy Rotation constraints would otherwise be
used for this purpose but cannot be used for various reasons. Basically, this
works in practically the same way that the Copy Rotation offsets work, including
the same weirdness that you'll get when trying to manually rotate these in the
3D viewport using "global" space manipulations ("local/normal" spaces though
still seem to work really nicely).
WARNING: this may potentially break old files with transform constraint setups
involving rotation outputs. Please check whether this causes any problems on old
files, and report back if there are any issues.
When you convert a grease pencil stroke to a polygon curve and look at the
vertices, the first and last vertex have weight = 0, but all others have a -NaN
value. This was caused by division by zero issues when minmax_weights[0] ==
minmax_weights[1].
that is that RNA does not support multiple inheritance so the actual node group subtypes (ShaderNodeGroup, CompositorNodeGroup, TextureNodeGroup) can not be derived from both the
ShaderNode/CompositorNode/TextureNode base types as well as a common NodeGroup type ... It is possible however to define node group types entirely in python which avoids the limitations of the RNA system
and is much more flexible, example for this will follow later.
* VERTEX_WEIGHT modifiers had quite a messy UI, improved the grouping of options a bit, so it is easier to scan through.
* Do not use abbreviations like "Rem" or "Dist" in the UI...
* Also small change for BEVEL modifier UI for more efficient space usage.
multiple importance sampling, so you can disable them for diffuse/glossy/transmission.
The Light Path node here is still weak and does not give this info. To make that
work we'd need to evaluate the shader multiple times which is slow and we can't
detect well enough when it is actually needed.
instead of sobol. So far one doesn't seem to be consistently better or worse than
the other for the same number of samples but more testing is needed.
The random number generator itself is slower than sobol for most number of samples,
except 16, 64, 256, .. because they can be computed faster. This can probably be
optimized, but we can do that when/if this actually turns out to be useful.
Paper this implementation is based on:
http://graphics.pixar.com/library/MultiJitteredSampling/
Also includes some refactoring of RNG code, fixing a Sobol correlation issue with
the first BSDF and < 16 samples, skipping some unneeded RNG calls and using a
simpler unit square to unit disk function.
NodeInternal as a base, which is a wrapper to expose the C callbacks as methods in bpy. Now these basic node types are also based on NodeInternal to ensure the full interface is available to py scripts.
In the process removed the unused NodeGroup register function, this doesn't work nicely anyway because it requires multiple inheritance which RNA doesn't support (so py node groups should be done
entirely in python in the future).
Caused by my recent normal calculation changes, added dependsOnNormals callback which was missing for ocean modifier (it assumed input normals were set).
