Is it possible to fine tune the model on the fly for any random image of a single person ?

zj-dong / AG3D

Official code release for ICCV2023 paper AG3D: Learning to Generate 3D Avatars from 2D Image Collections

https://zj-dong.github.io/AG3D/

244 stars 21 forks source link

Is it possible to fine tune the model on the fly for any random image of a single person ? #8

Open mayank64ce opened 11 months ago

zj-dong commented 10 months ago

Yes, you can optimize latent code or latent 3D tri-plane to fit our model to a single person. Since our training data is not so large, it may fail when these exists a large gap between training image and text image.