SUDO-AI-3D / zero123plus

Code repository for Zero123++: a Single Image to Consistent Multi-view Diffusion Base Model.
Apache License 2.0
1.72k stars 120 forks source link

Elevation of single input image and output images #97

Open JennySeidenschwarz opened 1 week ago

JennySeidenschwarz commented 1 week ago

Hi everyone!

Thanks for the cool work :) I have a question wrt to the elevation though: you said here that the model is trained using input elevation angles between -45 and 60 degrees absolute elevation and the output images are np.array([20, -10, 20, -10, 20, -10]). absolute elevation. However, elevation is very ambiguous for some objects. So it is super hard to define any kind of mapping between input and output elevations. Did you investigate on this further by any chance?

Best, Jenny

eliphatfs commented 1 week ago

Can you give me some examples where elevation is ambiguous? Since we live in a world with gravity, elevation works quite well.