2023-07-27 18:48:51
Stability AI has released Stable Diffusion XL 1.0, which it calls the “most advanced” release to date.
What we know
Joe Penna, Head of Applied Machine Learning at Stability AI, said Stable Diffusion XL 1.0 contains 3.5 billion parameters. The model can produce full 1 megapixel resolution images with multiple aspect ratios in seconds.
According to the creators, the algorithm provides “brighter” and “more accurate” colors, as well as better contrast, shadows and highlights than its predecessor.
Stable Diffusion XL 1.0 supports paint over, paint over and frame-to-frame indices. This means that users can give the model an input image and modify it.
According to Stability AI, the model includes complex, multi-part instructions given by short messages. Previous models required longer text prompts.
The developers have also trained Stable Diffusion XL 1.0 to generate clear and readable texts.
Stability AI recognized that the model might generate toxic and biased content. However, the developers have tried to adjust it to reduce the likelihood of generating such images.
Stable Diffusion XL 1.0 is distributed as open source on GitHub, in addition to the Stability API and consumer applications such as ClipDrop and DreamStudio.
Source : TechCrunch TechCrunch
1690509365
#Stability #releases #Stable #Diffusion #image #generation #model
