Open spthermo opened 6 years ago
My team's experience supports your claim. In this link ( https://we.tl/ODK9I5AIFy ) I enclose two mesh files. If corto is used to encode "input.ply", it encodes it but after decoding what is produced is a bad-indexed mesh "output.ply". We are highly interested in rectifying the situation and possibly contribute to it.
I have noticed that the number of vertices of the mesh changes after encoding (e.g. from 5700 to 5718 vertices). When the encoded mesh is noisy (e.g. person reconstruction from rgb-d sensors) the decoder cannot handle the difference in vertices. Has anyone faced the same problem?