We always see some interesting things coming out of digital graphics conference Siggraph Asia, and recently Microsoft Research showed off a piece of tech that can quickly learn your drawings and provide a template for future drawings and animations.
The tech doesn’t have a proper name yet, other than “Autocomplete hand drawn animations”. After taking one image as input, it can use that as a template to give you suggestions after you start a new image, and make adaptations on the fly. Want basically the same thing but with the head tilted a different way? Start drawing the head lines, and it’ll instantly recognise what you’re doing.
This can be great for animations, when you want very similar images in quick succession, just slightly moved. It can also detect patterns as you’re drawing them, and provide ghost lines for you to fill in. Use its template or don’t, it’s up to you — quick inputs can fill in the rest of the template for you, which could make drawing the same image over and over a much less time-consuming task.
It can even tie it all together in the form of an animation, as the video above shows.
There are no plans to bring it to market yet, nor is there available notes on the research behind it. Though, that could be a good thing — if the tech becomes widely circulated, we may see a lot more arguments about being a “tracer”.