It is a bug here? Index '[0]' seems wrong compared with orginal GS repo.

[HappyCoderBo]

depth-aware-3DGS/scene/gaussian_model.py

Line 134 in cfacf16

dist2 = torch.clamp_min(distCUDA2(torch.from_numpy(np.asarray(pcd.points)).float().cuda())[0], 0.0000001)

[HappyCoderBo]

which will cause the error 'IndexError: The shape of the mask [2108] at index 0 does not match the shape of the indexed tensor [1, 3] at index 0'

[JX-Github-web]

After removing [0], it can run, but the result is very poor, the depthloss is greater than 10, and the depth supervision is wrong.
Are your results normal?

这将导致错误 'IndexError： The shape of the mask [2108] at index 0 does not match match of the indexed tensor [1， 3] at index 0'

[HappyCoderBo]

After removing [0], it can run, but the result is very poor, the depthloss is greater than 10, and the depth supervision is wrong. Are your results normal?

这将导致错误 'IndexError： The shape of the mask [2108] at index 0 does not match match of the indexed tensor [1， 3] at index 0'

yes, my results are OK, but i use my own datas and depths predicted by DepthPro.