Question about the input single image

SUDO-AI-3D / zero123plus

Code repository for Zero123++: a Single Image to Consistent Multi-view Diffusion Base Model.

Apache License 2.0

1.71k stars 118 forks source link

Question about the input single image #8

Closed ingra14m closed 11 months ago

ingra14m commented 11 months ago

Hi, thank you for your outstanding work.

I have always had a question regarding the 'Single Image to 3D' tasks. I've noticed that almost all images input into these systems are in a front-facing and upright position. I'm uncertain if I would get equally perfect results when inputting images that are viewed from the bottom or upside down. I believe these latter cases could play a crucial role in many completion tasks.

eliphatfs commented 11 months ago

We only trained the model with input elevation angles between -45 and 60 degrees, so it would not be able to take views from extreme bottom. For most objects in the world, viewing from bottom gives very limited information about what the object is and how it looks like, so the problem would be a little ill-defined.

ingra14m commented 11 months ago

Thanks for your prompt reply!!