Classic-Generative-Model

Introduction

1. 本项目旨在以 Toy Demo😄😄的方式实现经典的 **AIGC⚡⚡**生成模型，避免下载复杂的数据集和需使用高端 GPU 的问题。
2. 通过直达模型🔭🔭的核心代码（配合论文阅读更佳），学习模型的核心思想，达到快速学习的目的。
3. 以下均为个人学习 AIGC 所阅读和编写的部分代码✨✨，欢迎指出问题和 Pull Request👯👯（PR）。
4. 以下均为个人学习 AIGC 所阅读的优秀博客📕📕和经典论文📕📕，感谢所有作者的贡献✨✨✨✨！！

Contents

1. A simple demo about DDPM in the DDPM folder.
2. A simple demo about GAN in the GAN folder.
3. A simple demo about VAE in the VAE folder.
4. A simple demo about VQVAE in the VQVAE folder.
5. A simple demo about PixelCNN and VQVAE in the PixelCNN_VQVAE folder.
6. A simple demo about Autoregressive Transformer and VQVAE in the AR_VQVAE folder.
7. A simple demo about Conditional Autoregressive Transformer in the Cond_AR_Transformer folder.
8. A simple demo about Calculating Motion FID in the Cal_Motion_FID folder.
9. A simple demo about Conditional Diffusion Using Classifier-Free Diffusion Guidance in the Cond_diffusion_Free_guide folder.
10. A simple demo about Using Diffuser in the Diffuser_pipeline folder.
11. A simple demo about Using LoRA based on the PEFT in the PEFT_LoRA folder.
12. A simple demo about MoE Model in the Simple_MoE folder.
13. A simple demo about DDIM Model in the Simple_DDIM folder.
14. A simple demo about UNet Model of Stable Diffusion in the SD_UNet folder.
15. A simple demo about U-ViT Demo in the Simple_U-DiT folder.
16. A simple demo about DDIM of Stable Diffusion in the SD_DDIM folder.
17. A simple demo about Inpainting of Stable Diffusion in the SD_Inpainting folder.
18. A simple demo about Classifer Guidance in the Simple_Classifer_Guide folder.
19. A simple demo about Diffusion Transformers (DiTs) in the Simple_DIT folder.

Blog Recommendation

• DDPM = 拆楼 + 建楼
• 一文带你看懂DDPM和DDIM
• 通俗理解GAN
• DiT详解
• 变分自编码器
• 变分自编码器
• 轻松理解 VQ-VAE
• 文生图模型之Stable Diffusion
• LoRA 在 Stable Diffusion 中的三种应用
• 扩散模型中的v-prediction
• 深入浅出完整解析ControlNet
• Stable Diffusion 原理介绍
• 一文读懂DDIM凭什么可以加速DDPM的采样效率
• 自回归图像生成代码阅读：VQ-GAN
• 详解VQGAN（一）| 结合离散化编码与Transformer的百万像素图像生成
• 保持ID的人像生成技术介绍：IP-Adaptor,PhotoMaker,InstantID
• Classifier Guidance 和 Classifier Free Guidance
• SD和Sora们背后的关键技术！一文搞懂所有 VAE 模型（4个AE+12个VAE原理汇总）
• AIGC专栏4——Stable Diffusion原理解析-inpaint修复图片为例
• AIGC-Stable Diffusion之Inpaint(图像修复)
• [论文理解] Classifier-free diffusion guidance
• Stable Video Diffusion 结构浅析与论文速览
• Stable Video Diffusion 源码解读 (Diffusers 版)
• AIGC-Stable Diffusion之VAE
• 深入浅出完整解析Stable Diffusion XL（SDXL）核心基础知识
• LoRA微调中是怎么冻结和加入AB矩阵的
• stable diffusion常用的LoRA、Dreambooth、Hypernetworks四大模型差异详解
• LoRA vs Dreambooth vs Textural Inversion vs Hypernetworks
• stable diffusion——Dreambooth原理与实践
• dreambooth原理
• Open-Sora全面开源模型参数和所有训练细节
• 论文阅读 | Smartbrush
• BrushNet —— 插件式图像修复inpainting 模型

Paper Recommendation

Highlights

• Rich Human Feedback for Text-to-Image Generation (CVPR 2024 best paper) [Paper] [Code]
• Generative Image Dynamics (CVPR 2024 best paper) [Paper] [Project]
• Scaling Rectified Flow Transformers for High-Resolution Image Synthesis (ICML 2024 best paper) [Paper] [Project]
• VideoPoet: A Large Language Model for Zero-Shot Video Generation (ICML 2024 best paper) [Paper] [Project]
• Discrete Diffusion Modeling by Estimating the Ratios of the Data Distribution (ICML 2024 best paper) [Paper]
• Visual Autoregressive Modeling: Scalable Image Generation via Next-Scale Prediction (NeurIPS 2024 best paper) [Paper] [Code]

Diffusion

• Denoising Diffusion Probabilistic Models [Paper] [Code]
• DENOISING DIFFUSION IMPLICIT MODELS [Paper] [Code]
• High-Resolution Image Synthesis with Latent Diffusion Models [Paper] [Code] [Code]
• Diffusion Models Beat GANs on Image Synthesis [Paper] [Code]
• Stable Video Diffusion: Scaling Latent Video Diffusion Models to Large Datasets [Paper] [Code]
• Align your Latents: High-Resolution Video Synthesis with Latent Diffusion Models [Paper]
• Elucidating the Design Space of Diffusion-Based Generative Models [Paper]

Autoregressive Model

• Autoregressive Image Generation without Vector Quantization [Paper]

GAN

• Generative Adversarial Nets [Paper] [Code]

VQVAE & VAE & VQ-GAN

• Neural Discrete Representation Learning [Paper]
• Auto-Encoding Variational Bayes [Paper]
• CV-VAE: A Compatible Video VAE for Latent Generative Video Models [Paper] [Code]
• Taming Transformers for High-Resolution Image Synthesis [Paper] [Code]

DIT

• Scalable Diffusion Models with Transformers [Paper] [Code]
• Hunyuan-DiT : A Powerful Multi-Resolution Diffusion Transformer with Fine-Grained Chinese Understanding [Paper] [Code]
• All are Worth Words: A ViT Backbone for Diffusion Models [Paper] [Code]

AI-generated image detection

• A Single Simple Patch is All You Need for AI-generated Image Detection [Paper] [Code] [Blog]
• PatchCraft: Exploring Texture Patch for Efficient AI-generated Image Detection [Paper] [Code] [Blog]
• GenDet: Towards Good Generalizations for AI-Generated Image Detection [Paper] [Code]

Project Recommendation

• stable-diffusion [Code]
• ControlNet [Code]
• AnimateDiff [Code]
• Open-Sora [Code]
• Open-Sora-Plan [Code]
• VAR [Code]
• IC-Light [Code]
• IP-Adapter [Code]
• LivePortrait [Code]
• zero123 [Code]
• VideoTetris [Code]
• Hunyuan3D [Code]
• LTX-Video [Code]
• HunyuanVideo [Code]

Thanks

Thanks to all the authors of the above blogs and papers!!