about inference speed

[WenjiaWang0312]

Hi author, thank you for your excellent work~

Have you tested the approximate speed of Easi3R?

[rover-xingyu]

Hi @WenjiaWang0312, thanks for your interest. The speed and GPU memory usage of Easi3R are similar to DUSt3R without flow loss and to MonST3R with flow loss. This is because the main computational cost comes from global alignment, which is the same for Easi3R and these models, while the extra cost from the second inference is minimal.