This project has moved to GitLab: https://gitlab.com/hard-diffusion/HardDiffusionBackend Inspired by EasyDiffusion
This works, kinda, expect issues.
Docs: https://billschumacher.github.io/HardDiffusion/
Moved to organization and split the monorepo
Backend (Django Rest Framework): https://github.com/HardDiffusion/HardDiffusionBackend
Frontend (Flutter): https://github.com/HardDiffusion/HardDiffusionFrontend
Rendering (Diffusers & Celery): https://github.com/HardDiffusion/HardDiffusionRenderer
New Flutter UI:
Generate:
Model Installation:
Status:
- Feature parity with EasyDiffusion, except installation.
- Distributed image processing ✅
- Use multiple models ✅
- Documentation ✅
- Update Documentation for multi repo setup.
- Tests
- Generating objects/characters with alpha background
- Object-like prompt? less NLP more {'background': 'prompt...', characters: [{...}]}
- UI Improvements
- Better renderer status
- Better error reporting ✅
- An efficient way to load and display historical results. ✅
- Custom VAE / Unets
- Textual Inversion inference
- Checkpoint Training
- Textual Inversion Training
- Prompt generation
- Multi-language prompts
- Prompt persistence
- Composable prompts
- Generate 3d models
- Export to blender
- Generate animations
- ControlNet (https://github.com/lllyasviel/ControlNet)
- Text-to-motion (https://github.com/Mael-zys/T2M-GPT)
- Realfusion (https://github.com/lukemelas/realfusion)
- Videos
See https://billschumacher.github.io/HardDiffusion/contributing.html
To inspect render performance first get the process ID then:
py-spy record -f speedscope -o out --pid {{ process_id }}
On Windows you can use tasklist after performance a single render it should be the python process with 4gb of memory or so.
On linux use ps aux | grep python
MIT with use restrictions, see LICENSE.
Painting Icon by Vick Romero, License is uncertain... Error Icon also had no clear license... https://dribbble.com/shots/3854786-Loading-Screen-1-5-Painting