Qualcomm AI Research has created an on-device version of Stable Diffusion, an artificial intelligence model that produces realistic images from textual descriptions in less than 100 seconds.
The usual deployment of Stable Diffusion takes place on cloud-based platforms, but the AI team at Qualcomm has managed to integrate it into Android smartphones by utilizing the Hexagon Processor of the Snapdragon 8 Gen 2 SoC.
This feat was accomplished by leveraging Qualcomm’s AI Stack, AI Model Efficiency Toolkit, and AI Engine, which facilitated the conversion of Stable Diffusion into an INT8-based version compatible with the Hexagon Processor.
Moreover, the complete process will occur within the device itself, as per Dr. Jilei Hou, Vice President of Engineering at Qualcomm, who asserts that this approach offers several advantages, including but not limited to privacy, security, rapidity, and reliability when using different network connections.
The VP further highlights the benefits of AI technology within the device, emphasizing its ability to enable various other applications, such as image manipulation, resolution enhancement, restyling, and many others.
The company made a way to create pictures using AI on a phone very quickly. The test phone did it in only 14.42 seconds.
This technology has many benefits on mobile, such as working without the internet, performing quickly, and keeping your information safe. However, to get the full effect of this technology, you will need to have the latest version of Snapdragon.
Related Stories: