DiscoArt provides a one-line Python API and CLI for generating Disco Diffusion images from text prompts. It requires a CUDA-enabled GPU and offers serving, result recovery, and configuration export features for integration into applications.
Project overview
The project reduces the setup complexity of Disco Diffusion to a one-line API or YAML-driven CLI, adding result recovery and serving endpoints for application integration.
Project type
Image & Vision · Prompt Engineering
Use cases
Image Processing
Deployment
Refer to project documentation
License
License pending
Best for
Developers and creators seeking an API-first or CLI-driven approach to generating Disco Diffusion images.
Key capabilities
Generates images from text prompts using Disco Diffusion in one line of code.
Provides `create` and `config` CLI commands to run generation tasks from a YAML configuration file.
Enables recovery of results from lost or early stopped sessions by pulling the session ID.
Runs DiscoArt as a service supporting gRPC, HTTP, websockets, and TLS.
Generates runnable Python code directly from a saved configuration.
Returns intermediate results as they become available via a `/result` endpoint.
Limitations and risks
Operation requires a CUDA-enabled GPU.
Parallel unblocked create requests can lead to Out of Memory (OOM) errors.
Getting started
Install the package via pip, configure CUDA-enabled PyTorch, and run the `create()` function or CLI command.
Evidence and sources
README: DiscoArt is an elegant way of creating compelling Disco Diffusion
README: DiscoArt has a modern & professional API with a beautiful codebase
README: Create compelling Disco Diffusion artworks in one line