Hacker News new | past | comments | ask | show | jobs | submit login

dreambooth retains higher fidelity as the model is finetuned, but to be honest I think textual inversion is actually more applicable as you can just add some embeddings to inject new knowledge into the model and not an whole new model just for a single concept (if you want to share it with others). Also I have not seen dreambooth being applied to replicate styles.



Consider applying for YC's Fall 2025 batch! Applications are open till Aug 4

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: