Qualcomm Stable Diffusion brings AI art generators to Android

Qualcomm’s Stable Diffusion model transforms text prompts to images on devices running on the Snapdragon 8 Gen 2 Mobile Platform

Cats in armor generated by Qualcomm Stable Diffusion
Qualcomm has provided an example of what their Qualcomm Stable Diffusion can do. These cats were generated with the prompt: "Super cute fluffy cat warrior in armor, photorealistic, 4K, ultra detailed, vray rendering, unreal engine."

Qualcomm has announced and demonstrated how the Qualcomm Stable Diffusion model uses text input to create photorealistic images in seconds. With this move, Qualcomm is taking on popular AI text-to-image art generators like Midjourney and giving Android users more talking points in the iOS vs Android debate.

Qualcomm says its goal in launching the Qualcomm Stable Diffusion model is to pave the way for AI text-to-image generators to overcome existing barriers and scale in the real world. To do this, Qualcomm believes the technology must no longer be confined to the cloud, hence its decision to design the Qualcomm Stable Diffusion model to run locally on smartphones powered by its Snapdragon 8 Gen 2 Mobile Platform.

After extensive research and iterating on existing text-to-image AI art technology, Qualcomm was able to shrink the technology using various tools developed by Qualcomm AI Research to create a model that offers increased performance and power efficiency without sacrificing the model’s accuracy. The research allowed the model to run on the Snapdragon 8 Gen 2 Mobile Platform.

What differentiates Qualcomm Stable Diffusion?

According to Qualcomm, the Qualcomm Stable Diffusion model takes advantage of software and hardware features such as the Qualcomm AI Engine framework and Qualcomm Hexagon Processor to offer users increased efficiency, better reliability, and lower cost compared to Qualcomm’s competitors.

The company claims that users can run the model on their devices without an internet connection, and use it to generate any picture they can imagine without compromising their privacy since the text prompts will never leave the device. Qualcomm emphasized user privacy as crucial to unlocking mainstream adoption by consumers and enterprise users.

Furthermore, Qualcomm announced that it would continue to advance its research because it believes the Stable Diffusion model has the potential to offer functionality beyond just text-to-image generation. The company predicts that in the not-too-distant future, the model could be used for image editing, painting, style transfer, and super-resolution photos.

Qualcomm says it is working on scaling the technology beyond smartphones so it can be deployed in laptops, XR headsets, and other devices powered by Qualcomm chipsets too.

World’s first on-device demonstration of Stable Diffusion on an Android phone



If you buy something from a GEEKSPIN link, we may earn a commission