Open malamutes opened 2 months ago
maybe it has something to do with the pipeline of convert.py, i'm trying to prepare a dataset from colmap instead, is it possible to pipe it through to train.py after automatic reconstruction in colmap? there were some issues with distortion but i think i can fix that in extras -> undistortion in colmap
I had a similar issue and I fixed it by using the same version of COLMAP that the authors use. Make sure you git clone version 3.8 from COLMAP
what is possibly the issue? these same set of images someone else ran on theirs and it worked beautifully but for mine it just outputs 2 images which are distorted