[14:48 Sat,10.June 2023 by Rudi Schmidts] |
Stable Diffusion is an open source AI image generator that enjoys great popularity precisely because of its openness. Thus, Stable Diffusion is basically freely available and can even be run locally on one&s own graphics card. However, the dataset trained so far was hardly subject to any restrictions, which leaves some copyright issues open.
;) Stable Diffusion XL However, as far as photorealism is concerned, Stable Diffusion has so far not been able to hold a candle to the best application for this: Because as long as you want to create deceptively real-looking photos, MIdjourney currently leads the field without a doubt. ;) Stable Diffusion XL But now, with Stable Diffusion XL, a revolutionary leap is perhaps in the offing. Not only at the request of affected artists, 80 million images are said to have been removed. Compared to the last version, the weights of the model are said to be more than twice as large with 2.3 billion parameters. Currently, Stability CEO Emad Mostaque announces on Twitter that one nevertheless has GPUs with 8GB memory in mind as  target system and Stability XL will be able to compress to this memory size. ;) Stable Diffusion XL Current beta versions are already available via the paid DreamStudio and users confirm that much higher quality results are now possible, especially in the area of photorealism. ;) Stable Diffusion XL Stable Diffusion XL is also said to be able to generate meaningful and readable text on images. What sounds trivial has so far been a major problem for generative AI models. One hundred percent reliability, however, should still be expected here - nevertheless, Stability XL should thus go down in history as the first generative AI model that is basically capable of text. That is, if no other usual suspect breaks this target before the official launch.  more infos at bei stability.aideutsche Version dieser Seite: Stable Diffusion XL: KI-Bilder werden fotorealistischer |
 |
|
|