Snapchat wants to speed up the response time for generating images with AI. The answer is a far faster model for building visual responses based on text queries.

Which I wouldn't have thought, had been a major barrier to adoption. Most generative AI tools currently take perhaps 30 seconds or so to generate such images, even on mobile devices. But Snap says that its new system can produce similar visualizations within less than two seconds - which, while not necessarily a huge game-changer, is an interesting development in the broader context of generative AI process.

As explained by Snap:

"SnapFusion cuts the model runtime from text input to image generation on mobile to under two seconds–the fastest time published to date by the academic community. Snap Research achieved this breakthrough by optimizing the network architecture and denoising process, making it incredibly efficient while maintaining image quality.". So now, it is possible to run the model to generate images based on text prompts and get crisp clear images back within seconds on mobile devices rather than minutes or hours as other research presents. One factor is an improved user experience, but Snap also notes that the new process may facilitate improved privacy: by limiting data sharing to third parties and reducing processing costs for developers.

Where Snap's research does include a few asterisks - including, most notably, that the majority of its experiments were conducted on an iPhone Pro 14, which, in Snap's own words 'has more computation power than many other phones'. As such, it's probably doubtful anything less than this is going to meet these speed benchmarks - but it'll still likely be quicker than current systems.
It's a fascinating experiment in itself, but it also indicates the future of a generative AI that will, eventually be able to respond to user cues in real-time, which may allow for a whole range of new usage options, such as real-time translation and more responsive creation.