How about fintune your base model on images of other resolution like 450*450

sstzal / DFRF

[ECCV2022] The implementation for "Learning Dynamic Facial Radiance Fields for Few-Shot Talking Head Synthesis".

MIT License

335 stars 40 forks source link

Open JunjieLl opened 1 year ago