Stable Diffusion XL 1.0 Is Now Available on Mobile


On the left: Stable Diffusion 1.5 / On the right: Stable Diffusion XL 1.0. The images are generated from identical text prompts via AI ARTA app

AIBY’s top-ranking AI-powered image-generating app AI ARTA introduced Stability AI’s SDXL 1.0 model to a global mobile audience.

MIAMI, FLORIDA, UNITED STATES, July 31, 2023/ — The release of Stable Diffusion XL (SDXL) 1.0 – the latest and most advanced version of Stability AI’s flagship text-to-image models – occurred on July 26, followed by an imminent debut on mobile executed by AI ARTA, the API’s official customer, who made the next generation model available for iOS and Android users worldwide on July 27.

Aiming to empower mobile users with easy and timely access to the most cutting-edge technology advancements in image generation, AI ARTA was the first mobile application to emerge with an upgraded model and its game-changing performance improvements. As noted by the AI ARTA development team, brief prompts are processed better, and the quality of generations regular users can get on their phones has grown significantly in all aspects, compositional, stylistic, and aesthetic, greatly enhancing the value of generated content.

According to Stability AI, SDXL 1.0 has one of the largest parameter counts of any open-source image model, boasting a 3.5B parameter base model and a 6.6B parameter model ensemble pipeline, while its predecessor was trained on 890M. The refinement in learning parameters enabled more accurate color reproduction, higher image contrast, more precise lighting and shadowing, and more subtle level of details than a base model. The dimensions of the generated output doubled, with 1024×1024 images being rendered now.

“We have strong evidence to expect remarkable quality boost of popular features based on Stable Diffusion: avatars, controlnet generations, inpainting/outpainting, image2image redrawing, QR code generations, video results, and much more – all of them promise to get impressively better”, says Ernest Piatrovich, Product Manager at AI ARTA.

“It’s worth noting, though, that SDXL 1.0 is more demanding on resources, Piatrovich adds, “which can drop the availability of generations for a standard amateur enthusiast. The waiting time on the external paid services side can also increase as even top servers may find it challenging to complete the generation of such high-quality images in less than 15 seconds. However, we have devised a plan to make this issue nearly imperceptible.”

AI ARTA is an AI-powered image-generating app that creates unique state-of-the-art pieces from text prompts, drawings, paintings, and photos. The app operates on Stability AI deep learning model Stable Diffusion. AI ARTA was ranked #2 in the App Store US Top Charts three months after its release date and currently holds the #3 position in Graphics and Design with 5M+ downloads worldwide.

Anastasia Kuchur

Source link


Please enter your comment!
Please enter your name here