In less than 15 seconds, Qualcomm demonstrated for the first time using Stable Diffusion to generate AI images on Android models.
CTOnews.com February 24 news, Qualcomm released a video on its official YouTube channel, the first successful use of Stable Diffusion on Android phones to generate AI images, the entire generation time does not exceed 15 seconds. Qualcomm said it was the first time in the world that the demo was on an Android device.
CTOnews.com Small Classroom: Stable Diffusion is a very popular base model, a text-to-image generative AI model capable of creating realistic images from any text input in tens of seconds.
Stable Diffusion has more than 1 billion parameters and currently runs primarily in the cloud. Qualcomm's demo uses Qualcomm AI Stack on Android devices to run Stable Diffusion on Android devices.
In the demo video, Qualcomm started with Hugging Face's FP32 version 1-5 open source model and optimized it to run on phones equipped with Snapdragon 8 Gen 2 mobile platform through quantization, compilation and hardware acceleration.