search

KU-CVLAB / 3DFuse

Official implementation of "Let 2D Diffusion Model Know 3D-Consistency for Robust Text-to-3D Generation"
711 stars 42 forks source link

bad case #21

Closed No360201 closed 1 year ago

No360201 commented 1 year ago

when i set prompt is " cute small corgi" and "a ladybug",why the result is this

https://user-images.githubusercontent.com/57511687/236725037-a2da8675-a8cd-496a-a22f-e40e7173c2a4.mp4

https://user-images.githubusercontent.com/57511687/236725063-f2c46181-5fae-4b3d-aa12-05942700c7b5.mp4

i changed semantic_coding.py (image[~mask] = [255., 255., 255.]) to image[~mask[:,:,3]] = [255., 255., 255.].

j0seo commented 1 year ago

Hello, it appears that there has been error propagation of Point-E during the point cloud inference process. Although this framework can address some errors in point clouds, it struggles with point clouds that are completely geometrically incorrect in inference, and we leaved this as future work. Please refer to #11 for more information on this matter.

No360201 commented 1 year ago

Hello, it appears that there has been error propagation of Point-E during the point cloud inference process. Although this framework can address some errors in point clouds, it struggles with point clouds that are completely geometrically incorrect in inference, and we leaved this as future work. Please refer to #11 for more information on this matter.

you mean if we provide better point cloud ,we can get a right result

j0seo commented 1 year ago

Yes, generally, there is a tendency toward that.