Weights were calculated using tessellation data, giving slightly uneven weighting.
Now only use tessellation for ray-cast but weight the influences from the original polygons.
Also cache arrays from derived-mesh, they we're called each intersection.
Now the conversion code uses mesh element mapping to speed up lookups.
Gives really nice speed improvement here, but the cost is higher memory
usage during refiner construction.
On the dragon scene here topology refiner construction time goes down
from 5 seconds to around 0.01.
It's possible to reduce the memory footprint by allocating mapping in
stages (don't allocate all of them at once, but do it on demand only
and free them after they're not needed anymore).
Caused by own rBff3d535bc2a6309 - since we now only write the exact amount of layers
needed to store saved customdata, we have to adjust CustomData->maxlayer too.
Otherwise, on next read, customdata code believes it has more layers allocated than
actual number.
Issue reported by Campbell over IRC, thanks.
The error was "ValueError: Function <function normal_at_I0D at 0x7f2aad1feb70>
has keyword-only arguments or annotations, use getfullargspec() API which can
support them", and was first seen in eeeb845d33e81afbc8ed127e6ab4ae7b18472a54
That was basically not an issue with interpolation, but rather missing wrapping
options and periodic wrapping was always used.
It's still a bit questionable why certain graphics cards were doing clamping in
the file from the report, that's not something what is expected to happen from
the settings of textures being passed to GPU. In any case this issue i still
didn't manage to reproduce on any of the available GPUs, might be something
related on driver glitch or so.
In any case CPU now should behave just fine, rest of the issues we'll need to be
able to reproduce first.
Currently only two mappings are supported by API, which is Repeat (old behavior)
and new Clip behavior. Internally this extension is being converted to periodic
flag which was already supported but wasn't exposed.
There's no support for OpenCL yet because of the way how we pack images into a
single texture.
Those settings are not exposed to UI or anywhere else and there should be no
functional changes so far.
Works totally similar to camera motion blur and majority of the changes are
related on just passing extra arguments to sync() functions.
Couple of things still to look into:
- Motion pass will not include motion caused by the zoom.
- Only perspective cameras are supported currently.
- Motion is being interpolated on projected coordinates, which might give
different results from constructing projection matrix from interpolated
field of view.
This could be good enough for us, but we need to consider improving this
at some point.
Reviewers: juicyfruit, dingto
Reviewed By: dingto
Differential Revision: https://developer.blender.org/D1383
I suspect code here can be cleaned up but for now try this.
Alternatively we can check for errors around buffer allocation
but this needs bigger changes.
Finding a specific F-Curve is often needed in Python, and usually
consists of a construct like:
```
[fcurve
for fcurve in ob.animation_data.action.fcurves
if fcurve.data_path == "location"][1]
```
This can now be written as
`ob.animation_data.action.fcurves.find('location', 1)`
This new function `Action.fcurves.find()` is still O(N) in the number
of FCurves in the Action, but at least it allows us to remove
boiler-plate code. It is also faster than the Python equivalent, as
only the found F-Curve is converted to Python.
Reviewers: campbellbarton, aligorith
Reviewed By: aligorith
Differential Revision: https://developer.blender.org/D1427
properly.
This should fix failure to use vertex arrays in OSX with high
polycounts.
Note this will not suffice as a fix when we move to VBOs exclusively
(GL 3+), we'll have to think of some way to separate huge meshes to many
VBOs.
Issue is rather well explained in T45471: our current customdata writing code easily generates several different blocks in blend file with same 'old' address. This is bad, because those addresses are used as 'uid' during reading process (it kind of work in Blender's own reading process, by mere luck mostly, but breaks the file specs).
Solution (suggested by Campbell, thanks) implemented by this patch is to avoid duplicating everything, and instead just overwrite what we needs to skip some cdlayers on write:
* the CustomData's `totlayer` number;
* the CustomData's `layers` array of CustomDataLayer (keeping its original address using the `writestruct_at_address` helper).
New design allows us to get completely rid of the no_free flag stuff in `write_customdata()`.
Note that this implies written data is **not** directly valid from Blend PoV, since its written typemap does not match written layers (this is not an issue because typemap is rebuilt on read anyway - and it's easy to fix this if really needed).
Also, the backward compatibility saving of mface data remains an issue here, see comment in code.
Reviewers: sergey, campbellbarton
Projects: #bf_blender
Maniphest Tasks: T45471
Differential Revision: https://developer.blender.org/D1425
This commit contains all the remained parts needed for initial integration of
OpenSubdiv into Blender's subdivision surface code. Includes both GPU and CPU
backends which works in the following way:
- When SubSurf modifier is the last in the modifiers stack then GPU pipeline
of OpenSubdiv is used, making viewport performance as fast as possible.
This also requires graphscard with GLSL 1.5 support. If this requirement is
not met, then no GPU pipeline is used at all.
- If SubSurf is not a last modifier or if DerivesMesh is being evaluated for
rendering then CPU limit evaluation API from OpenSubdiv is used. This only
replaces the legacy evaluation code from CCGSubSurf_legacy, but keeps CCG
structures exactly the same as they used to be for ages now.
This integration is fully covered with ifdef and not enabled by default
because there are several TODOs to be solved first:
- Face varying data interpolation is not really cleanly implemented for GPU
in OpenSubdiv 3.0. It is also not implemented for limit evaluation API.
This basically means we'll have really hard time supporting UVs.
- Limit evaluation only works with adaptivly subdivided meshes so far, which
basically means all the points of CCG are pushed to the limit. This gives
different result from old code.
- There are some serious optimizations possible on the topology refiner
creation, which would speed up initial OpenSubdiv mesh creation.
- There are some hardcoded asumptions in the GPU and DerivedMesh areas which
could be generalized.
That's something where Antony and Campbell can help, making it so the code
is structured in a way which is reusable by all planned viewport projects.
- There are also some workarounds in the dependency graph to make sure OpenGL
buffers are only freed from the main thread.
Those who'll be wanting to make experiments with this code should grab dev
branch (NOT master) from
https://github.com/Nazg-Gul/OpenSubdiv/tree/dev
There are some patches applied in there which we're working on on getting
into upstream.
Those files are still not in use (SCons will tyr to compile new CCGSubSurf files
but no code will be in use at all because those new files are fully wrapped by
ifdef WITH_OPENSUBDIV check).
