【论文目录】2013-2024年AI图像生成技术30篇+经典论文—— 图像生成Diffusion模型演化进程——扩散or自回归 ？

曾小蛙-CSDN博客 整理发布  : https://blog.csdn.net/imwaters/article/details/144969523

10年图像生成的发展脉络  

1. 2013.12 VAE:  Auto-Encoding Variational Bayes
2. 2014.06 GAN:  Generative Adversarial Networks （生成对抗网络--上一代图像生成架构）
3. 2017.06 Transformer :  Attention Is All You Need
4. 2017.11 VQ-VAE : Neural Discrete Representation Learning
5. 2018.12 StyleGAN: (系列模型，GANs时代高峰，开启高清人像生成)A Style-Based Generator Architecture for Generative Adversarial NetworksGitHub - NVlabs/stylegan: StyleGAN - Official TensorFlow Implementation
6. 2020.06 DDPM: Denoising Diffusion Probabilistic Models 
7. 2020.10 DDIM: Denoising Diffusion Implicit Models
8. 2021.02  (自回归)DALLE: Zero-Shot Text-to-Image Generation
9. 2021.03 CLIP: Learning Transferable Visual Models From Natural Language Supervision
10. 2021.05 Diffusion Models Beat GANs on Image Synthesis guided-diffusion: Diffusion Models Beat GANs on Image Synthesishttps://github.com/openai/guided-diffusion
11. 2021.05 (自回归) CogView:Mastering Text-to-Image Generation via Transformershttps://github.com/THUDM/CogView （可中文题词）
12. 2021.08 SDEdit:                  (使用随机微分方程引导图像合成和编辑)  Image Synthesis and Editing with Stochastic Differential Equations
13. 2021.12.GLIDE: openai开发的模型Towards Photorealistic Image Generation and Editing with Text-Guided Diffusion Models
14. 2021.12  LDMs  High-Resolution Image Synthesis with Latent Diffusion Models GitHub - CompVis/stable-diffusion: A latent text-to-image diffusion model
15. 2022.04 DALLE-2:  Hierarchical Text-Conditional Image Generation with CLIP Latents
16. 2022.04 CogView2 : （对标DALLE-2）论文：Faster and Better Text-to-Image Generation via Hierarchical Transformers https://github.com/THUDM/CogView2
17. 2022.05 Imagen: Photorealistic Text-to-Image Diffusion Models with Deep Language Understanding
18. 2022.07 CFG: Classifier-Free Diffusion Guidance

19. 2022.08 Stable Diffusion 开源 ： 论文是LDMs：GitHub - CompVis/stable-diffusion: A latent text-to-image diffusion model

20. 2022.10 Flow mactch: Flow Matching for Generative Modeling (SD3部分理论来源)
21. 2022.11 SD2 https://huggingface.co/stabilityai/stable-diffusion-2
22. 2022.12  DIT:  Scalable Diffusion Models with Transformers (DiT)https://github.com/facebookresearch/DiT   (SORA的基础架构)
23. 2023.02 ControlNet:  Adding Conditional Control to Text-to-Image Diffusion Models GitHub - lllyasviel/ControlNet: Let us control diffusion models!
24.  2023.07 SDXL: Improving Latent Diffuon Models for High-Resolution Image Synthesishttps://huggingface.co/stabilityai/stable-diffusion-xl-base-1.0
25. 2023.09 DALLE-3:  Improving Image Generation with Better Captions
26. 2023.10 LCM: Latent Consistency Models Latent Consistency Models: Synthesizing High-Resolution Images with Few-step Inference
27. 2023.11 AnyText （中文图片生成） Multilingual Visual Text Generation And Editing GitHub - tyxsspa/AnyText: Official implementation code of the paper <AnyText: Multilingual Visual Text Generation And Editing>
28. 2023.12 Imagen2 未开源无论文 Imagen 2 - Google DeepMind
29. 2024.03 CogView3  Finer and Faster Text-to-Image Generation via Relay Diffusion
30. 2024.03 SD3: Scaling Rectified Flow Transformers for High-Resolution Image Synthesis
31. 2024.04  Hyper-SD: (少步数采样的sota) Trajectory Segmented Consistency Model for Efficient Image Synthesis
32. 2024.04 VAR: Visual Autoregressive Modeling: Scalable Image Generation via Next-Scale https://github.com/FoundationVision/VAR
33. 2024.05  腾讯HunyuanDiT: A Powerful Multi-ResolutionDiffusionTransformer with Fine-Grained ChineseUnderstanding 
34. 2024.07  快手Kolors  :  （可中文题词）Effective Training of Diffusion Model forPhotorealistic Text-to-Image Synthesis​​​​​​  
35. 2024.08 Flux.1 : 论文是SD3, 最强开源文生图模型GitHub - black-forest-labs/flux: Official inference repo for FLUX.1 models

主要参考

万字长文分享快手 Kolors 可图大模型应用实践_kolors模型-CSDN博客

​​

小何同学梳理的发展脉络

【扩散生成模型串讲】一个视频梳理Diffusion模型演化进程，16个扩散相关生成模型，20+篇相关论文串讲。_哔哩哔哩_bilibili

​​