Thanks for your amazing work and sharing of the project! I use data captured by phone for reconstruction,
the camera poses were estimated by colmap, but the reconstruction results were not good
I thinked the results is rendered by incorrect camera poses, so I use the hloc for sfm, and write a script to convert the hloc-sfm output files to cams_meta.npy. But the reconstruction results including using the ngp_fox dataset were still not good. Can you help me analyze the problem?
Thanks for your amazing work and sharing of the project! I use data captured by phone for reconstruction, the camera poses were estimated by colmap, but the reconstruction results were not good I thinked the results is rendered by incorrect camera poses, so I use the hloc for sfm, and write a script to convert the hloc-sfm output files to cams_meta.npy. But the reconstruction results including using the ngp_fox dataset were still not good. Can you help me analyze the problem?