Georgia Gkioxari

If you need to report an issue, you need to provide data and a compact code snippet that reproduces the error.

It's my understanding that we are deprecating `projects/nerf` with the new implicitron library

@bottler @davnov134 Can you point Sergey to the code that reproduces NeRF for LLFF and Lego?

Hi @wzic Thanks for bringing this up! It does indeed sound interesting! This op is currently not in our immediate roadmap. However, if more users request it we will include...

Ugh! Unfortunately, the issue is numerical. Let me explain! There is an `EPS` in the code which performs various checks between box coordinates. https://github.com/facebookresearch/pytorch3d/blob/276c9a8acbfa715f5802f26ec9f4141bde26ecb4/pytorch3d/csrc/iou_box3d/iou_utils.cuh#L15 Now your boxes are **really** close...

Yes. The distance in the box coordinates is close to `EPS`. So then in the code, when do some `if-else` checks (I don't want to go into implementation details!) the...

For full disclosure, this doesn't happen generally for `EPS`- close boxes. This numerical instability is particularly triggered by your example. For example, below the boxes are also * very close*...

Ah interesting! Could you perhaps provide a few more (3-4) test cases so that I add them to our [test function](https://github.com/facebookresearch/pytorch3d/blob/main/tests/test_iou_box3d.py) to make sure the solution works for more of...

@psteeves The issue is now fixed by https://github.com/facebookresearch/pytorch3d/commit/1bfe6bf20a1de877cc623d11c2eeed8c7091ae90 I added your test cases in our [tests](https://github.com/facebookresearch/pytorch3d/blob/main/tests/test_iou_box3d.py). Some of them I couldn't quite reproduce even with the original code. For example,...

The `dists` attribute is the 2D distance of each pixel to the corresponding 3D point (indexed by the appropriate fragments attribute). Not sure what the error is here? What would...