Qualcomm has recently unveiled an exciting new development in mobile technology, showcasing its AI capabilities with the latest iteration of its image generator, Stable Diffusion. The company has claimed that this is the fastest-ever deployment of the software on a smartphone, with a demo video showing it generating a 512 x 512-pixel image in under 15 seconds. While Qualcomm has not revealed the specific phone used, it has confirmed that it was powered by its flagship Snapdragon 8 Gen 2 chipset. The software has also undergone a range of custom optimizations to achieve optimal performance.

Qualcomm

This is an impressive feat, particularly as generating an image in this way on a decent laptop can take minutes, let alone seconds on a smartphone. Most apps offering AI image generation do all their processing in the cloud, rather than on the mobile device itself. Dall.E is an example of one such AI image generation software that runs all of its processing threads on the cloud, which is why it generates images with the speed it does. Qualcomm’s achievement marks a significant step forward in the productization of AI, with the potential to offer greater convenience, cost savings, and privacy.

While Qualcomm has claimed that this is a speed record, some developers have pointed out that they have already achieved similar results with Stable Diffusion running on Android devices. However, this doesn’t take away anything from the impressive nature of Qualcomm’s achievement. It’s important to remember that getting big AI models running locally on mobile devices is a complex task that requires significant computing power. Plus, privacy is highly enhanced since no data is being sent over the internet. By showcasing this technology, Qualcomm has demonstrated its expertise in this field and opened up exciting possibilities for the future of mobile AI.

RELATED:

(Via)