About the comparison experiment setup in papper

[zzg-zzg]

I have a question about the comparison experiment setup. When you compare GPS-GS with other methods, such as 3DGS, do you use all surrounding images as 3DGS input, or just two images for one person?

[ShunyuanZheng]

When compared with feed-forward methods (i.e., Floren, ENeRF, and IBRNet), we use two input views.
However, as for 3DGS, we use all 8 source views as input.