2023 · Today Stability AI, together with its multimodal AI research lab DeepFloyd, announced the research release of DeepFloyd IF, a powerful text-to-image cascaded. DeepFloyd IF uses a modular approach to AI image generation that is very similar to Google's Imagen. DeepFloyd generative art uses a frozen T5 XXL which is a Large Language Model (LLM), a text-to-image diffusion model that generates a 64x64px image, then two super-resolution diffusion models that upscale the image to 256x256px and then 1024x1024px.