Closed lazyDamon closed 9 months ago
Since VolRecon is able to render depth with good quality on ETH3D, it should have good generalization ability. Since your mesh looks very coarse, I think the problem may come from tsdf_fusion. First, I think you can check whether rendered depth maps look reasonable. Second, I think you need to tune 'voxel_size' in tsdf_fusion.sh. On DTU, voxel_size is set to 1.5 (the metric is millimeter). You need to adjust it based on your scale and metric (meter or millimeter).
Hi, thank you for your great work! On dtu datasets, it looks good. I try to extract the mesh on custom datasets. First, I use
script/eval_general.sh
to rendering image and depth. Then I try to runtsdf_fusion.sh
to get the reconstructed meshes. Finally, I changed the image shape and view list to my own datasets inevaluation/clean_mesh.py
. But results are badly. Here are some of my results, hope you can help me, thank you!