Introducing Flux
How does Flux work? The new image generation AI that rivals Midjourney
What is Flux
Official Technical Blog : Flux is a transformer-powered flow model at scale.
Flux is a new AI image generation model developed by Black Forest Labs. It represents a significant advancement in AI-generated art, utilizing a “hybrid architecture” that combines transformer and diffusion techniques, scaled up to 12 billion parameters. The model offers state-of-the-art performance image generation with top of the line prompt following, visual quality, image detail and output diversity.
The model has been proposed by Black Forest Labs, a new research lab founded the original creators of the Stable Diffusion models for image and video generation (Stable Diffusion XL, Stable Video Diffusion, Rectified Flow Transformers), and Adversarial Diffusion Distillation for ultra-fast, real-time image synthesis.
The most recent works from these researchers are:
- Fast High-Resolution Image Synthesis with Latent Adversarial Diffusion Distillation (March, 2024)
- Scaling Rectified Flow Transformers for High-Resolution Image Synthesis (ICML, 2024)
How does it work?
All public FLUX.1 models are based on a hybrid architecture of multimodal [1] and parallel diffusion transformer [1, 2] blocks and scaled to 12B parameters. We improve over previous state-of-the-art diffusion models by building on flow matching [3], a general and conceptually simple method for training generative models, which includes diffusion as a special case. In addition, we increase model performance and improve hardware efficiency by incorporating rotary positional embeddings [4] and parallel attention layers [5].