Anttwo / SuGaR

[CVPR 2024] Official PyTorch implementation of SuGaR: Surface-Aligned Gaussian Splatting for Efficient 3D Mesh Reconstruction and High-Quality Mesh Rendering
https://anttwo.github.io/sugar/
Other
2.36k stars 182 forks source link

After editing the 3D GS training results, use SuGaR for optimization #208

Open seulqxq opened 3 months ago

seulqxq commented 3 months ago

Very good work, I got great results using SuGaR!

I would like to ask if I can only optimize the part I want. For example, I use SuperSplat and other tools to edit the ply scene generated by 3D GS, and then use SuGaR method to optimize and extract the Mesh.

This is the scene trained by 3D GS: image

This is the scene after I edited it, which is the part of the mesh that I want to extract: image

Can I use edited scenarios (point_cloud_sp.ply) instead of the original 3D GS-generated scenarios (point_cloud.ply) before using SugGaR for training?

I tried, but the results were not very good: image

image