Urgent Issues

[Nancyllx]

Hi, thanks for sharing this great work!

We are working on a very urgent project and just want to raise two problems we encountered:

1. When running the neural texture infer stage (command: infer_texture.py --input_path=data/inference_samples/azure_04 --texture_out_dir=data/textures/azure_04) with our own data (pkl, json, original images and segmented images), the output is still a texture from your dataset, rather than matching with our data input. Do you know what the problem is? Could you please give us some advice since we have been troubled by this problem for a long time but have been unable to solve it.
2. The dimension of the body pose parameter in your SMPL-X file (xxx.pkl) is 63, while the dimension in our pkl file (obtained through Smplify-X project) is 32. We manually changed our dimension to 63 (in order to align with yours) - will this operation bring any negative impact? Besides, is it convenient to reveal what changes you have made?

Above are the problems we are struggling with. Hope to get your earliest reply. Thank you for your help and support! For any further contact, you can email 2987291309@qq.com :)