Closed ansj11 closed 4 months ago
I can't see the uploaded video.
sorry, the video maybe too big to upload, I only upload 2 screenshots.
You can change the sig and lap_weight to get a smooth mesh. There is no need to use all the images for reconstruction, and the varying lighting can affect the results. you can select like 20 images for reconstruction and ignore those images where the person is completely sideways.
thank you for your response! I will have a try!
I have 500 images from iPhone with colmap camera pose, and I use colmap2mvsnet.py to get pair.txt and points.pkl, but the optimized mesh looks not good. Do you know why?
Uploading 屏幕录制2024-05-31 10.38.17.mov…