Solved

[muses0229]

[Zzlongjuanfeng]

yeah, we have discussed the multi-face problem in our paper. This issue primarily arises from the inconsistency of multi-view texture images sampled from the pre-trained 2D diffusion model, as it is not explicitly trained on multi-view datasets.

Some new models (zero123, EscherNet) consider camera pose during the diffusion process, which may alleviate the multi-face problem.