SDXL-Turbo is a fast generative text-to-image model that can synthesize photorealistic images from a text prompt in a single network evaluation. A real-time demo is available here: http://clipdrop.co/stable-diffusion-turbo

Key Takeaways:

  • SDXL Turbo achieves state-of-the-art performance with a new distillation technology, enabling single-step image generation with unprecedented quality, reducing the required step count from 50 to just one.

  • See our research paper for specific technical details regarding the model’s new distillation technique that leverages a combination of adversarial training and score distillation.

  • Download the model weights and code on Hugging Face, currently being released under a non-commercial research license that permits personal, non-commercial use.

  • Test SDXL Turbo on Stability AI’s image editing platform Clipdrop, with a beta demonstration of the real-time text-to-image generation capabilities

Model weights and code: https://huggingface.co/stabilityai/sdxl-turbo

Demo: https://clipdrop.co/stable-diffusion-turbo

Paper: https://stability.ai/research/stability-ai-adversarial-diffusion-distillation

Blogpost: https://stability.ai/news/stability-ai-sdxl-turbo

  • Even_Adder@lemmy.dbzer0.comOP
    link
    fedilink
    English
    arrow-up
    1
    ·
    1 year ago

    I realized too late that this was being released under a non-commercial research license. Maybe they’ll open source it soon.

    • simple@lemm.ee
      link
      fedilink
      English
      arrow-up
      2
      ·
      1 year ago

      The blogpost says “It’s important to note that SDXL Turbo is not yet intended for commercial use.” so I think that implies that a commercial license is coming later.

        • simple@lemm.ee
          link
          fedilink
          English
          arrow-up
          2
          ·
          1 year ago

          Darn. At least commercial use isn’t super expensive all things considered. Makes sense they want to cash in on its use from companies, and thankfully the model is still technically open to use.

        • keepthepace@slrpnk.net
          link
          fedilink
          English
          arrow-up
          1
          ·
          1 year ago

          Understandable but sad. They are now at the bleeding edge and not playing catchup anymore. They can cash out hard. That means that for now, this advanced model is a dead end for open source as we wont be able to improve it as we were in the last iterations.

          “non-commercial license” is not open source but I would not mind if this became the standard for the cutting edge in the industry as we try to figure out a business model to make true open source work.