Dreambooth-Stable-Diffusion and hyperdreambooth
These are competitors offering alternative approaches to fine-tuning Stable Diffusion for personalized image generation: the first uses standard Dreambooth fine-tuning while the second replaces the full model fine-tuning with HyperNetwork adaptation for faster training.
About Dreambooth-Stable-Diffusion
XavierXiao/Dreambooth-Stable-Diffusion
Implementation of Dreambooth (https://arxiv.org/abs/2208.12242) with Stable Diffusion
Fine-tunes the entire diffusion model's U-Net weights (rather than just embeddings) using paired subject images and class-level regularization images to prevent overfitting. Leverages gradient checkpointing and the Stable Diffusion v1 architecture, requiring a rare token identifier and synthetic or real regularization images during training to maintain model generalization across semantic variations.
About hyperdreambooth
JiauZhang/hyperdreambooth
Implementation of HyperDreamBooth: HyperNetworks for Fast Personalization of Text-to-Image Models
Related comparisons
Scores updated daily from GitHub, PyPI, and npm data. How scores work