Your Galaxy S23 might be powerful enough to quickly generate AI images
AI images can now be generated quickly on Android phones. In a recent demo video, Qualcomm was able to generate a 512 x 512 pixel image of a cat in armor using AI-image-generator Stable Diffusion version 1.5 on an Android phone. What’s more, it finished in just under 14.42 seconds, which is pretty close to what a PC can do with the technology.
To achieve this, the company’s engineers used a tool called “QuantifyWithout delving into it here, they basically took the desktop version of Stable Diffusion, scaled it down, and optimized it for mobile. According to Qualcomm, the process “improved performance [while] return [saving] By allowing the model to “operate more efficiently thus”[consuming] Less memory bandwidth. ”
It’s unclear what mobile device was used in the demo.However, we do know that it is powered by the Snapdragon 8 Gen 2 mobile platform, which is the same as the Samsung Galaxy S23. It is worth mentioning that Other developers have managed to make Stable Diffusion exist Sony Xperia 5 II, but it takes about an hour to generate the image on that phone due to weak hardware. The main takeaway from what Qualcomm has achieved is its speed and what it means for users in the future.
Generating the Future of Artificial Intelligence
The thing about generative AI models is that they require a lot of computing power to work. Developer of Stable DiffusionFor example, a computer with NVIDIA hardware and at least 6.9 GB of VRAM (especially video RAM) is recommended for creating images. Smartphones don’t come with VRAM because manufacturers opt for regular RAM to run the software on the device. But now that power differential is starting to recede, so we could see this technology appear on more mobile devices.
Hou Jilei, vice president of engineering at Qualcomm, told venture capital Lessons learned from optimizing Stable Diffusion for smartphones can be applied to other devices running on corporate hardware. Hou singles out laptops and XR headsets as potential future homes for generative AI.As for what they can do, there are many applications such as “image editing [plus] in-painting,” the process by which artificial intelligence fills in missing parts in generated artwork. The company also hopes that all of this can be done without requiring an internet connection to the Stable Diffusion cloud. Cool as that might be, it’s not yet It’s unclear if or when the technology will be widely available on phones running Qualcomm hardware.
In addition to generating images, developers have also created writing AI to help write content. Be sure to check out our most recent list of the best AI writers of 2023—Jasper is among our favorites.