Wonbong Jang (Won)
Wonbong Jang (Won)
Hello @Seasandwpy , From my experience, the pose is optimized first then shape/texture latent vectors are optimized later according to the roughly estimated pose. If the network cannot find the...
Hello @Kulbear Shape/Texture code, as well as camera pose, are optimized simultaneously. What I meant in the previous reply was that even though all three are optimized simultaneously, the model...
I will try as you suggest! Thank you for your quick response and sorry for the late reply.
I think that the new version works much better - great work!!
You may need to modify 'render_bkgd' for the custom dataset. Pose may not be correct as well - if you use colmap, then it follows opencv - not opengl.