Qualcomm has demonstrated the power for Android phones to generate AI pictures inside seconds utilizing easy textual content prompts. The chipmaker was capable of create a 512 x 512 pixel picture of a warrior cat carrying armor utilizing an Android cellphone and Steady Diffusion, a preferred text-to-image generative AI mannequin.

Qualcomm says it utilized its AI stack to run Steady Diffusion 1.5 on the cellphone, which wasn’t linked to the web. Primarily, the cellphone carried out all AI operations essential to generate the picture on system. In a video shared by the corporate, the AI picture of the Cat is generated in just below 15 seconds.

Steady Diffusion created the picture utilizing the immediate: “Tremendous cute fluffy cat warrior in armor, photorealistic, 4K, extremely detailed, vray rendering, unreal engine.”

The cellphone that generated the AI picture inside seconds was utilizing the Qualcomm Snapdragon 8 Gen 2 chipset discovered on flagships just like the Galaxy S23 series. The Steady Diffusion mannequin used was optimized for “quantization, compilation, and {hardware} acceleration” earlier than testing.

This isn’t the primary occasion of Steady Diffusion producing a picture on an Android cellphone. As The Verge factors out, a developer named Ivon Huang beforehand used Steady Diffusion on a Sony Xperia 5 II powered by a Qualcomm Snapdragon 865 and 8GB of RAM. Nevertheless, it took Huang an hour to generate a 512 x 512 picture. So Qualcomm’s method is actually a lot quicker than what the developer used.

Having AI fashions like these working on Android telephones can save price for builders, remove the necessity for a cellular connection, and likewise defend person privateness as the info stays in your cellphone.

There’s no phrase from Qualcomm about when prospects can anticipate these fashions to run broadly on flagship Android telephones. The corporate will demo the tech at MWC 2023, so we’d hear extra about it within the coming days.