research for diffusion model
- On the Importance of Noise Scheduling for Diffusion Models
TL; DR: same noise level will have different influence on different resolution. - Common Diffusion Noise Schedules and Sample Steps are Flawed
TL; DR: terminal snr is not zero so information leak. - Perception Prioritized Training of Diffusion Models
- LoRA: Low-Rank Adaptation of Large Language Models
- Controlling Text-to-Image Diffusion by Orthogonal Finetuning
- DoRA: Weight-Decomposed Low-Rank Adaptation
- LoRA+: Efficient Low Rank Adaptation of Large Models
- Scalable Diffusion Models with Transformers
TL; DR: aka DiT
- Latent Consistency Models: Synthesizing High-Resolution Images with Few-Step Inference
- Adversarial Diffusion Distillation
- SDXL-Lightning: Progressive Adversarial Diffusion Distillation
- EchoMimic [too slow]
- Face0: Instantaneously Conditioning a Text-to-Image Model on a Face
- InstantID: Zero-shot Identity-Preserving Generation in Seconds
- PhotoMaker: Customizing Realistic Human Photos via Stacked ID Embedding
- IDAdapter: Learning Mixed Features for Tuning-Free Personalization of Text-to-Image Models
- Face2Diffusion for Fast and Editable Face Personalization
- Towards a Simultaneous and Granular Identity-Expression Control in Personalized Face Generation