Question about the resolution of training images on the Mipnerf-360 dataset.

city-super / Scaffold-GS

[CVPR 2024 Highlight] Scaffold-GS: Structured 3D Gaussians for View-Adaptive Rendering

https://city-super.github.io/scaffold-gs

Other

701 stars 59 forks source link

Question about the resolution of training images on the Mipnerf-360 dataset. #63

Open TY424 opened 2 months ago

TY424 commented 2 months ago

Hi and first of all thanks for your great work! The training image resolution of 3DGS is 2 times downsampled (indoor) and 4 times downsampled (outdoor). The training script 'train_mip360.sh' for Scaffold-GS does not seem to have downsampling parameters. What is the training resolution in the paper?

inspirelt commented 2 months ago

We use the default settings as shown by the code: rescaling large resolution to 1.6k.

TY424 commented 2 months ago

Hello, I have another question. The outdoor scene metrics in Tables 6-8 are lower than those of 3DGS. The visualization results of these scenarios were not presented in the paper. We retrained the model, but the results were not very good. The training parameters are python /GIT/Scaffold-GS/train.py --eval -s /GIT/Scaffold-GS/data/mipnerf360/bicycle --lod 0 --gpu -1 --voxel_size 0.001 --update_init_factor 16 --appearance_dim 0 --ratio 1 --iterations 30_000 -m /SCGS/bicycle.

Test image: 00001 00006

Is it convenient to display official results？

TY424 commented 2 months ago

Is there a problem with my parameter settings?

inspirelt commented 2 months ago

We use unified parameter configuration for all mipnerf360 scenes for simplicity. For the outdoor scene with tiny and thin objects like leaf and grass, a smaller voxel size should be better. Feel free to try and welcome to share the results.