Stability AI has announced Stable Diffusion 3, a “next generation” image synthesis model. It “uses the work of its predecessors to generate detailed, multi-subject images with improved quality and accuracy from text descriptions.”
Stability says the Stable Diffusion 3 family of models (which takes text descriptions called “hints” and turns them into corresponding images) has between 800 million and 8 billion parameters. This range allows you to run different versions of the model locally on different devices` from smartphones to servers. The number of parameters roughly corresponds to the model's capabilities in terms of how much detail it can generate. Larger models also require more VRAM on graphics accelerators to run, reports ArsTechnica.
Stability has been creating advanced AI image generation models since 2022: Stable Diffusion 1.4, 1.5, 2.0, 2.1, XL, XL Turbo, and now 3. The company has made a name for itself as a more open alternative to proprietary image synthesis models , such as DALL-E 3 from OpenAI, although not without controversy regarding the use of copyrighted training data, bias and potential for abuse. Stable Diffusion models can be run locally and fine-tuned to change the results.
Stability CEO Emad Mostak wrote on X:
This uses a new type of diffusion transformer (similar to the Sora diffusion transformer) combined with flux matching and other improvements. It takes advantage of transformer enhancements and can not only scale further, but also accept multimodal inputs.
Some notes:
— This uses a new type of diffusion transformer (similar to Sora) combined with flow matching and other improvements.
— This takes advantage of transformer improvements & can not only scale further but accept multimodal inputs..
— More technical details soon— Emad (@EMostaque) February 22, 2024
Sound designer course at gamedevi. For 29 borrowings you will get yourself a portfolio, software and hard skills and you can easily start in gamedev. Get to know more
Stable Diffusion 3 also uses “flow matching,” a technique for creating AI models that can generate images by learning to transition smoothly from random noise to a structured image. It does this without having to model each step of the process, instead focusing on the overall direction or flow that the creation of the image should follow.
Comparison of the results of OpenAI DALL-E 3 and Stable Diffusion 3 with the hint: “Night photo of a sports car with the inscription SD3 on the side, the car on racing track at high speed, a huge road sign with the word Faster written on it.”
Stable Diffusion 3 is not widely available, but Stability says once testing is complete it will be free to download and run locally.
The competition for ITS authors continues. Write an article about the development of games, gaming and gaming devices and win a professional gaming wheel Logitech G923 Racing Wheel, or one of the low-profile gaming keyboards Logitech G815 LIGHTSYNC RGB Mechanical Gaming Keyboard!