ForMyCat
commented
1 month ago If this is the case, you can try playing with SDS hyperparameters, or sampling looking-down views (instead of the random sampling in my code) to apply the SDS loss.
Closed ansj11 closed 1 month ago
ForMyCat
commented
1 month ago If this is the case, you can try playing with SDS hyperparameters, or sampling looking-down views (instead of the random sampling in my code) to apply the SDS loss.
Your work is impressive and I have some questions. I am training SparseGS with 32 horizontal views of a car, and I want use this method to get better looking-down views. But the results are not so good, do you have any suggestions? Here are my images and results.