Thanks for your brilliant work! Apparently this method is focused on sparse view synthesis. But I tried several datasets but found the results were not met with my expectation. So I considered using more images, Can this method also work in dense views? Also, I noticed that the dust3r pipeline significantly slowed down when using too many views (For ex. 300 views) as the image pairs generated will be n^2 where n is the image num. Looking forward to your reply
Thanks for your brilliant work! Apparently this method is focused on sparse view synthesis. But I tried several datasets but found the results were not met with my expectation. So I considered using more images, Can this method also work in dense views? Also, I noticed that the dust3r pipeline significantly slowed down when using too many views (For ex. 300 views) as the image pairs generated will be n^2 where n is the image num. Looking forward to your reply