Quenstion about table

[ms98x]

Hello, thank you very much for your work. I would like to ask if the PSNR indicator in the table is obtained by averaging the fine_psnr_ori_full values ​​obtained from training in each scene? The average value I calculated is slightly different from the one in the paper.

[cwchenwang]

Actually I forget about this. I think I calculate PSNR, SSIM and LPIPS afterwards. How much is the difference?

[ms98x]

I calculated the PSNR of blenderX2(100x100) and LLFFX2 datasets respectively. LLFF is 0.29 higher and Blender is 0.6 lower. So I'm not sure if you calculated it using the value of fine_psnr_ori_full. By the way, I encountered the following problem when training the BlenderX4 dataset, can you tell me how to solve it?

[cwchenwang]

I think it's good to calculate the metrics afterwards. Could you please tell me what script you used to train the blender x4 data?

[ms98x]

scripts/train_blender_downX.sh
I just modified the downscale.

[cwchenwang]

You should also change W and H to 400.

[ms98x]

Therefore, the width and height corresponding to blenderx4 (200X200) in the paper are
800X800? For blender(100X100)X4, I should change the width and height to 400x400 in the scripts.Can I understand it this way?

[cwchenwang]

Exactly.