Hello,
Thank you for sharing the code. I could not reproduce the reported values for novel view synthesis with ./scripts/metrics.sh, and I would like to ask if there’s any adjustment that should be made to the code. I conducted the reproduction on an H100 GPU, and here’s the result for subject 306 (mine vs reported):
SSIM : 0.9616672396659851 vs 0.975
PSNR : 32.85559844970703 vs 36.28
LPIPS: 0.03651179000735283 vs 0.035
DEPTH ERR: 0.886204496116002mm
Thank you!
Hello,
Thank you for sharing the code. I could not reproduce the reported values for novel view synthesis with ./scripts/metrics.sh, and I would like to ask if there’s any adjustment that should be made to the code. I conducted the reproduction on an H100 GPU, and here’s the result for subject 306 (mine vs reported):
SSIM : 0.9616672396659851 vs 0.975
PSNR : 32.85559844970703 vs 36.28
LPIPS: 0.03651179000735283 vs 0.035
DEPTH ERR: 0.886204496116002mm
Thank you!