Pretty soon there won’t be anything that artificial intelligence can’t do when it comes to manipulating photos and videos. We’ve told you about AI tech that can spin images out of whole cloth, change the weather in images, and even add new speech to extant video.
It’s pretty mind blowing and, as we’ve also said many times before, the next frontier in photo editing. Today’s story is no less cool or ground breaking than the others and probably some more immediate application than the other stories about AI that we have covered (you can check them out by clicking here).
This technology takes a figure in a still image and generates a full 3D, animated model based upon that. Researchers from the University of Washington and Facebook discussed their findings in a paper titled “Photo Wake-Up: 3D Character Animation from a Single Photo.”
Basically, it works like most AI applications in that it detects the human figure in the photo and then builds a 3D model around that silhouette. It then applies the image’s features to the model. It uses what PetaPixel likens to Photoshop’s content aware fill to make sure that any gaps in the data can be handled so that nothing looks weird or unnatural – such as filling in details not shown in the original image.
The University of Washington’s Ira Kemelmacher-Shlizerman said of the research, “The big challenge here is that the input is only from a single camera position, so part of the person is invisible…Our work combines technical advancement on an open problem in the field with artistic creative visualization.”
You can watch a video discussing the team’s research over on YouTube by clicking here.
As always, we’d love to know your comments – be sure to leave them below.
Don’t forget to check out our other news articles on Light Stalking by clicking here.