facebookresearch / audio2photoreal

Code and dataset for photorealistic Codec Avatars driven from audio
Other
2.66k stars 250 forks source link

Novel view #34

Closed ggfresh closed 7 months ago

ggfresh commented 8 months ago

Awesome project! Where could I set the camera params for rendering novel view?

evonneng commented 7 months ago

Hi, thanks!! For the camera params, you could change the K and Rt matrix that is getting passed into the renderer.

So for instance, I am just loading in some default camera parameters as per this line here: https://github.com/facebookresearch/audio2photoreal/blob/548aeeb2057465045ca1568d65ea059cea633d80/visualize/render_codes.py#L89

And passing in the parameters to the model here https://github.com/facebookresearch/audio2photoreal/blob/548aeeb2057465045ca1568d65ea059cea633d80/visualize/render_codes.py#L121

This is where these things get ingested. https://github.com/facebookresearch/audio2photoreal/blob/548aeeb2057465045ca1568d65ea059cea633d80/visualize/ca_body/models/mesh_vae_drivable.py#L283-L285

Since different parts of the body decoder use either the K or the Rt, you will need to set both. Hope this helps!

alexanderrichard commented 7 months ago

Closing this issue as it seems resolved. Please reopen if you have more questions.