Request for Scaffold-GS training parameters on MatrixCity block_small (PSNR≈29 in paper) #81
Description
Hello, thank you for your impressive work on Octree-GS.
I am currently trying to reproduce the MatrixCity block_small results reported in the paper, where the model trained with Scaffold-GS achieves around 29 PSNR.
However, despite multiple trials with different parameter settings, my reproduced results are still far below the reported performance.
Could you please share the complete training configuration (or the corresponding configuration file, script, or command line arguments) used for the Scaffold-GS setup on the MatrixCity block_small dataset?
It would be very helpful for understanding how this configuration contributes to the reported PSNR.
If it is more convenient, you could also send it to my email: [email protected]
.
Thank you very much for your time and for sharing your excellent work with the community.