Creating generative AI on users’ computers seemed far-fetched when ChatGPT, Dalle, Midjourney, and other large models were released less than two years ago. But thanks to model improvements, it’s become relatively easy to run a chatbot, for example.
Creating images is a heavier task than creating text, but even there things work. The solutions that have been found so far have mostly been trivial and require the user to be relatively knowledgeable, but now AMD has launched beguilea program that uses the Stable Diffusion algorithm to generate images based on text or graphic descriptions.
The program is simpler than most “AI on your computer” solutions. The user does not need to download any models or make any settings. The generation parameters can be changed just like in the online version of Stable Diffusion, for example, but this is not necessary. Once the images are ready, the program uses the XDNA Supersolution feature to scale them up so that they can be used for more purposes.
Amuse is optimized for AMD hardware and initially only works on PCs with Radeon RX 7000 series graphics cards, Ryzen AI 300 (Strix Point) chipset, or Ryzen 8040 (Hawk Point) family chipset. The RX 6000 series and Ryzen 7040 family do not meet the requirements, although the latter is almost identical to Hawk Point.
More Stories
EA President Talks New Dragon Age: 'A Return to What Made Bioware Great'
She thought she had bought a phone – she was shocked by its contents
Rumor: Lots of AI in Google's Pixel 10 and 11 cameras