Openjourney is a fine-tuned version of Stable Diffusion v1.5, created by PromptHero, designed to replicate the distinctive visual style of Midjourney. It generates high-quality, imaginative text-to-image outputs when prompted with the phrase “mdjrny-v4 style.” Built on top of Stable Diffusion, it maintains the same architecture and parameters but introduces stylistic tuning through curated training on Midjourney-style images. Openjourney is compatible with Hugging Face’s diffusers library and can be exported to ONNX, MPS, and JAX/FLAX formats. It supports torch float16 precision for efficient GPU inference and produces results that lean toward fantasy, surrealism, and visually complex compositions. The model is available for free use via Hugging Face Spaces, and additional resources like prompt collections and a LoRA version are also provided. It is ideal for artists, designers, and AI enthusiasts seeking creative image generation with a Midjourney-inspired aesthetic.
Features
- Fine-tuned on Midjourney-style images for artistic output
- Based on Stable Diffusion v1.5 architecture
- Requires “mdjrny-v4 style” in prompts for best results
- Supports inference with diffusers in PyTorch
- Compatible with ONNX, MPS, and JAX/FLAX export
- Uses torch float16 for fast GPU rendering
- Freely accessible via Hugging Face Spaces
- Includes sample prompts and resources for creative exploration