diffusion-study

Custom implementation of iterative image diffusion process

Build & Run docker

• docker build -t sdiffusion . - to build docker
• To run a container, use:

docker run -d -it --init \
--gpus=all \
--ipc=host \
--volume="$PWD:/app" \
--volume="/home/:/hhome" \
--volume="/usr/local/cuda:/usr/local/cuda" \
--publish="5555:5555" \
--publish="5556:5556" \
sdiffusion bash

Run jupyter server

• jupyter lab --no-browser --ip 0.0.0.0 --port 5555 --allow-root

Useful materials:

• Yannic Kilcher | DDPM - Diffusion Models Beat GANs on Image Synthesis (Machine Learning Research Paper Explained)
• Jeremy Howard | Lesson 9: Deep Learning Foundations to Stable Diffusion, 2022)
• https://github.com/fastai/diffusion-nbs - Tools and Notebooks from fast.ai
• From Autoencoder to Beta-VAE
• What are Diffusion Models?
• Tools and Resources for AI Art
• Private Local Waifu Diffusion colab
• Stable Diffusion Tutorial Part 1: Run Dreambooth in Gradient Notebooks
• Stable Diffusion Tutorial Part 2: Using Textual Inversion Embeddings to gain substantial control over your generated images
• Generating images with Stable Diffusion
• Stable Diffusion Models
• DreamFusion: Text-to-3D using 2D Diffusion
• High-Resolution Image Synthesis with Latent Diffusion Models | GitHub
• Denoising Diffusion Probabilistic Models
• Improved Denoising Diffusion Probabilistic Models
• DiffRF: Rendering-guided 3D Radiance Field Diffusion
• 3D Neural Field Generation using Triplane Diffusion
• Hugging Face Diffusion Models Course
• Diffusion samplers easy to use implementations

---

Datasets

• DDPM(Denoising Diffusion Probabilistic Models) was trained on MNIST and Anime Face Dataset