News for hackers. Stuff that matters.

A comment on DreamFusion: Text-to-3D using 2D Diffusion. Parent comment

Re:DreamFusion: Text-to-3D using 2D Diffusion by ShamelessC on Thursday September 29, 2022 @10:12PM (#33030064)
The model is trained on discretized photons (pixels). Doing so may have encouraged the model to learn features related to depth.
[ Parent | on HN ]

There are no comments on this comment yet.