Gul Varol

Thank you for your questions. We’ve addressed them below, please let us know if anything is unclear: (1) BOBSL vs BSL1K - Although BOBSL and BSL1K are constructed in a...

BSL-1K will not be released, sorry for the outdated repository. We have released BOBSL instead by reproducing all the papers where we had used BSL-1K.

What do you mean by 'my image'? These textures are captured from 3D scans of people which are then registered to SMPL model to create this UV map.

Sorry, generating texture maps is out of scope of this repository. You might be interested in "Dense Pose Transfer" ECCV 2018 by Neverova et al.

Sorry that I currently don't have time to check this. Unless you need this conversion for particular reason, you can also generate the point cloud from the SMPL parameters.

Feel free to open a pull request. I can no longer provide assistance on versions unfortunately. For the neutral model, please see: https://github.com/gulvarol/smplpytorch/issues/3#issuecomment-553190211

It was not removed, the original SMPL release did not contain a neutral model, it was developed afterwards in the linked `up` repository above.

Random clipping is selecting a random starting frame from the sequence, which is quite standard nowadays. The implementation is pretty clear: https://github.com/gulvarol/ltc/blob/master/donkey.lua#L42 Multiscale cropping is a simple augmentation as well:...

You mean for training or testing? For training, we ensure to choose a random clip with enough frames to avoid this: https://github.com/gulvarol/ltc/blob/master/donkey.lua#L42 For testing, according to your example, we take...

No we sample a random starting frame between 1 and 191, so we use all frames. math.ceil(torch.uniform(1e-2, N-loadSize[2]+1)) N = 210 loadSize[2] = 20 The interval becomes [1, 191]