miaosheng

After update the nerfstudio 2.0, I found that the training speed of nerfstudio2.0 is unstable compared with nerfstudio1.x. I want to know the reason and how to fix it to...

is there any scripts to inference only with rgb images input based on your pretrained model

4

Thank you for your contribution! I want to use this method as backbone, is there any scripts to inference only with rgb images input based on your pretrained model?

Mono depth Loss should be implemented with patchsampler, in the sdfstudio i find it implemented with random pixelsampler?

1

Thanks for your contributation! In the monosdf paper it describe the scale invariant loss function should be used for individually per image batch. But when i run the progarm, i...

Thanks for your contribution! Now I want to test this method on sparse urban scene. I used the pretrained mipnerf360 checkpoint because it is suited for the unbounded scene. Now...

Why normalize the intrinsics and extrinsics in the covert_dtu.py?

4

For the DTU dataset, I noticed your code has normalized the camera intrinsics in the convert.py? As a result, the pose and intrinsics in the code are hard to understand...

For large-scale scenes, particularly in the forward-moving street scenes, some 3D Gaussians may lie outside the current camera frustum. A better way to save memory and accelerate the speed is...

@torch_compile() for function get_viewmat() cause the bad training results.

1

**Describe the bug** I've been testing SplatFacto on my device and encountered a peculiar issue. When I follow the original code and apply `@torch_compile()`, my results end up in local...

Why Translation normalization have a huge impact on the rendering result?

3

Hi, I'm training the Mvsplat to reconstruct the street scene, and I find a question: * when i normalize the extrinsics (eg. translation ) using the following code, i can...

why fix the fx and fy when center-crop the image patch in the code?

2

hello. when you apply center-crop for the image, I think you should fix the `cx` and `cy` for the image intrinsics. But in your implementation, you fix the `fx` and...