[HN Gopher] 3D Novel View Synthesis with Diffusion Models
       ___________________________________________________________________
        
       3D Novel View Synthesis with Diffusion Models
        
       Author : dougabug
       Score  : 36 points
       Date   : 2022-10-04 19:55 UTC (3 hours ago)
        
 (HTM) web link (3d-diffusion.github.io)
 (TXT) w3m dump (3d-diffusion.github.io)
        
       | dr_dshiv wrote:
       | It seems like this be used to create multiple views for fine
       | tuning Stable Diffusion (textual inversion), from a single image.
        
       | dougabug wrote:
       | This approach is interesting in that it applies image-to-image
       | diffusion modeling to autoregressively generate 3D consistent
       | novel views, starting with even a single reference 2D image.
       | Unlike some other approaches, a NeRF is not needed as an
       | intermediate representation.
        
       | muschellij2 wrote:
       | Soon to be the Face Back APP!
        
       | mlajtos wrote:
       | Ok, NeRFs were a distraction then.
        
       | oifjsidjf wrote:
       | >> In order to maximize the reproducibility of our results, we
       | provide code in JAX (Bradbury et al., 2018) for our proposed
       | X-UNet neural architecture from Section 2.3
       | 
       | Nice.
       | 
       | OpenAI shitting their pants even more.
        
         | astrange wrote:
         | Oh, OpenAI does more or less release that much. People don't
         | have issues implementing the models from their papers.
         | 
         | What they don't do is release the actual models and datasets,
         | and it's very expensive to retrain those.
        
       ___________________________________________________________________
       (page generated 2022-10-04 23:00 UTC)