PSNR about Mip-NeRF360
wenkaier opened this issue · comments
The original paper (3DGS) is 27.21. In your paper, it is 28.69. What processing did you do?
The original 3dgs is averaged on all 9 scenes of mip360. However, the data of scene treehill
and flowers
were not public, previously. So our reported result is averaged on the rest 7 scenes.
scenes = ["bicycle", "bonsai", "counter", "flowers", "garden", "stump", "treehill", "kitchen", "room"]
factors = [4, 2, 2, 4, 4, 4, 4, 2, 2]
The above experiment setup is for mip-splatting on the Mip-NeRF360 dataset. I would like to ask if the table results provided in your article are also under this experimental setup?
No. We did not set the resolution factor, details can be found in train.sh
and train_mip360.sh
.