**Stability AI** has announced the launch of **SDXL 0.9**, a significant advancement in the Stable Diffusion text-to-image suite of models. This new version offers massively improved image and composition detail, opening up creative possibilities for generative AI imagery in various fields such as films, television, music, design, and industrial use.
The key driver behind the improved composition in **SDXL 0.9** is its significant increase in parameter count, making it one of the largest open-source image models available. It utilizes a 3.5B parameter base model and a 6.6B parameter model ensemble pipeline, resulting in hyper-realistic imagery with finer details and a higher resolution of 1024×1024.
**SDXL 0.9** is accessible through ClipDrop with API access coming soon, and it can be run on a modern consumer GPU. The model offers functionalities beyond text prompting, including image-to-image prompting, inpainting, and outpainting.
## Supplemental Information โน๏ธ
Stability AI has introduced **SDXL 0.9**, a major advancement in their Stable Diffusion text-to-image models. This release significantly improves the level of image and composition detail compared to its predecessor. SDXL 0.9 expands the creative potential of generative AI imagery, making it suitable for various industries such as entertainment, design, and manufacturing.
The model’s parameter count plays a crucial role in its enhanced composition capabilities. With a 3.5B parameter base model and a 6.6B parameter model ensemble pipeline, SDXL 0.9 achieves hyper-realistic images with finer details and a resolution of 1024×1024. It utilizes two CLIP models, including OpenCLIP ViT-G/14, one of the largest OpenCLIP models to date.
SDXL 0.9 offers functionalities like image-to-image prompting, inpainting, and outpainting, expanding its applications beyond text-based inputs. It can be accessed through ClipDrop, and an API will be available soon. Despite its advanced capabilities, SDXL 0.9 can be run on a modern consumer GPU.
Researchers will have access to SDXL 0.9 for a limited period to provide feedback and refine the model before its general open release. The model is currently intended for research purposes only and is subject to a non-commercial, research-only license.
### ELI45 ๐
Stability AI has released SDXL 0.9, an advanced text-to-image model that significantly improves image and composition detail. It can generate hyper-realistic images for various applications like films, TV, music, design, and industry. The model has a large parameter count and can run on a modern consumer GPU. It offers additional features beyond text prompting and is currently available for research purposes.
#### Source ๐: https://stability.ai/blog/sdxl-09-stable-diffusion?format=amp