Your Android phone might soon be able to generate AI images in seconds

Qualcomm has demonstrated the ability for Android phones to generate AI images within seconds using simple text prompts. The chipmaker was able to create a 512 x 512 pixel image of a warrior cat wearing armor using an Android phone and Stable Diffusion, a popular text-to-image generative AI model.

Qualcomm says it utilized its AI stack to run Stable Diffusion 1.5 on the phone, which wasn’t connected to the internet. Essentially, the phone performed all AI operations necessary to generate the image on device. In a video shared by the company, the AI image of the Cat is generated in just under 15 seconds.

Stable Diffusion created the image using the prompt: “Super cute fluffy cat warrior in armor, photorealistic, 4K, ultra detailed, vray rendering, unreal engine.”

The phone that generated the AI image within seconds was using the Qualcomm Snapdragon 8 Gen 2 chipset found on flagships like the Galaxy S23 series. The Stable Diffusion model used was optimized for “quantization, compilation, and hardware acceleration” before testing.

This isn’t the first instance of Stable Diffusion generating an image on an Android phone. As The Verge points out, a developer named Ivon Huang previously used Stable Diffusion on a Sony Xperia 5 II powered by a Qualcomm Snapdragon 865 and 8GB of RAM. However, it took Huang an hour to generate a 512 x 512 image. So Qualcomm’s technique is certainly much faster than what the developer used.

Having AI models like these running on Android phones can save cost for developers, eliminate the need for a mobile connection, and also protect user privacy as the data remains on your phone.

There’s no word from Qualcomm about when customers can expect these models to run widely on flagship Android phones. The company will demo the tech at MWC 2023, so we might hear more about it in the coming days.