Hello there,
We have a volumetric capture studio that uses multi-depth cameras to reconstruct 4d sequence data. It can capture full-body movements or focus on the facial expressions at 30 FPS.
We are currently trying to use the 4D sequence data as an input to train the MLDeformer. We would start with the facial expressions. Could you provide us with some guidelines on this process? Additionally, are you interested in training for full-body movements?