Qianli Ma

By the way I forgot to mention that the raw scans are by default not colored either. You can specify that you need colored scans when requestion. Sorry I deleted...

Hi, the model is subject-specific, e.g. the pre-trained skirt model will always generate skirt results in the given pose. So yes you need to retrain on the new data. Please...

Hi, training on a single pose won't make the model generalizable to unseen poses (as in this case the model doesn't have the knowledge how the shape should deform as...

Hi! You can take a look at [this discussion](https://github.com/qianlim/POP/issues/5#issuecomment-962938549) in the POP repo and see if it resolves your question. Also, on the CAPE dataset website, we also provide the...

The SCALE model is a point-based model; the output from the code is always a point set. To get a mesh, you can run mesh reconstruction algorithms such as Poisson...

Hello, the local-to-global transformation here means the transformation of each point between the canonical pose and the target pose. The full transformation should be a 4x4 matrix but here we...

Hello, yes, that's correct. It depends on the resolution of the UV map. The pixels on the UV 'islands' correspond to points on the body surface. The 'faceid' here means...

Hi, sorry for the late reply. For each valid pixel on the UV map, it corresponds to a point on the body surface, and the `bary_coords_smpl_uv` gives the barycentric coordinates...

Hi! Yes the CAPE dataset contains the "single mesh alignment" from ClothCap, but the segmented meshes are unfortunately not provided. Thanks for understanding.

Hi, the test set RGB images used in ICON test set is being prepared and will be available on the CAPE website soon. Stay tuned!