Stability AI released Stable Diffusion XL (SDXL), its updated AI image generator that "excels at photorealism."
U.K.-based
Stability says SDXK beta targets enterprise clients and is available
through the company's API and DreamStudio, its generative art tool.
- SDXL
reportedly offers improved photorealism, image composition, legible
text, and face and hand generation over its predecessor, the Stable
Diffusion 2.1 open-source model.
- Creators can also generate descriptive images based on simpler prompts, according to Stability.
- After trying out SDXL, Fast Company reported that its "photorealism goes way past the uncanny valley and sits firmly on the cusp of an Everest of believability."
- SDXL
was trained on 2.3 billion parameters versus 900 million for Stable
Diffusion V2.1, according to Stability VP of Engineering Scott Draves.
- Notably, Stable Diffusion is one of the models available in Amazon Web Services' new Bedrock suite of AI tools.
- Stability AI brought in a $100M investment last year in a round led by Coatue, Lightspeed Venture Partners, and O'Shaughnessy Ventures LLC.
- Semafor recently reported
that Stability has burned through "a significant chunk" of the cash.
Two anonymous venture investors expressed doubts about taking part in
another funding round that could bring Stability's valuation to $4B, up
from $1B last October.
