Cycles : Update to 5.1.0#6885
Conversation
4ad8b04 to
64a6cce
Compare
|
Thanks @boberfly, I've updated this to use the new I've added 73ee78c to work around the crash in our tests by preventing the principled_bsdf shader from requesting |
|
On my phone atm but I saw https://projects.blender.org/blender/blender/pulls/156693 which I think didn't make it into 5.1 standalone I think, I'll do a test with this and see. I'll look for more related patches. |
|
Ok I remember a bit more how this works - Blender hands off a In there to test this and a shuffle primitive variables node to copy P to Pref and this prevents the crashing. I'll test out a fix by moving that call. |
|
GafferHQ/dependencies#300 try this one out @murraystevenson I think this fixed it for me. |
73ee78c to
bb1dae7
Compare
Thanks boberfly, I've updated this to |
johnhaddon
left a comment
There was a problem hiding this comment.
Thanks @murraystevenson and @boberfly for pushing this along. I don't know what has caused it, but I'm seeing some issues rendering UVs, with messages like this one popping up a lot :
WARNING : IECoreCyles::GeometryAlgo::convertPrimitiveVariable : Primitive variable "uv" has size 96 but Cycles allocated size 0.
Any ideas?
| for( const auto &it : mesh->variables ) | ||
| { | ||
| if( it.second.interpolation == PrimitiveVariable::Uniform && it.first == "N" ) |
There was a problem hiding this comment.
Why are we doing linear search rather than mesh->variables.find()?
Scratch that - seems I just had a stale build. |
53b3ec0 to
c31be17
Compare
…r cycles_kernel_cpu still needs to link.
Cycles now outputs normalized normals, so we can no longer directly compare the result with our non-normalized primitive variable.
Now that Cycles has support for FaceVarying normals, we can preserve custom Uniform normals by resampling them to FaceVarying.
This removes the workaround added in 916f860. Cycles 5.2 will improve support for even numbers of samples, and we've backported that PR to Cycles 5.1 in GafferDependencies. https://projects.blender.org/blender/blender/pulls/157076
c31be17 to
98f1233
Compare
|
Thanks for the update Murray - I have squashed and merged. |
Generally describe what this PR will do, and why it is needed
Related issues
Dependencies
Breaking changes
Checklist