Andrew Luo

So technically [mitsuba2](https://github.com/mitsuba-renderer/mitsuba2) natively supports gradients for pytorch tensors. However there are many functions lacking (batched rendering, multi-gpu, [differentiation w.r.t. vertices](https://github.com/mitsuba-renderer/mitsuba2/pull/157) etc.) that make it not realistic to use for...

I wonder if another possible alternative would be to regularize the mesh itself to ensure that the transformations (weakly) preserve some metric of the mesh itself.

I see this in a few models in ShapeNetv2 as well, along side with other artifacts (missing texture, inverted faces). This looks like z-fighting, and if the dataset is only...

It would be very useful to see the full source of the project. Tagging @vacancy @chuangg 😀

Also using ManifoldPlus on shapenet as part of a SDF generation pipeline. Generally I see few artifacts at `--depth 9`, however there are still problems with complex objects like `cars`,...

Actually fixed the bug. DIB-R was missing a check, fixed it in my private version.

I believe it had to do with the gradients causing the texture values to go negative. I recall doing the following: 1. Sigmoid applied to vertex color (significantly slowed convergence)...

Just built sparse/scatter/cluster/spline-conv/geometric with PyTorch master, CUDA 11, cudnn8, GCC 9.3, MAGMA 2.5.2. Encountered no issues either compiling or running.

I would like to bump this request.

It would be nice to see this code merged!