A new algorithm has been developed by scientists that could transfer the signature styles of acclaimed photographers to your cellphone photos or selfies.
YiChang Shih, a Massachusetts Institute of Technology (MIT) graduate student said that "Style transfer" is an interesting g area of graphics research, but standard style-transfer techniques tend not to work well with close-ups of faces.
Shih and colleagues at Adobe and the University of Virginia have managed to perform a "local transfer."
They have used off-the-shelf face recognition software to first identify a portrait in the desired style and then used characteristics similar to those of the photo to be modified.
"We then find a dense correspondence - like eyes to eyes, beard to beard, skin to skin - and do this local transfer," Shih said.
Shih said that one consequence of local transfer is that the technique of researchers works much better with video than its predeces sors.
Shih, however, added that even local transfer cannot make modified photos look fully natural. So a new feature was added to the algorithm, which is called "multiscale matching." "Human faces consist of textures of different scales," Shih said.
He added that people want the small scale - which corresponds to face pores and hairs - to be similar, but they also want the large scale to be similar like nose, mouth, lighting.
via Science - Google News http://ift.tt/RV7V1h
Put the internet to work for you.
No comments:
Post a Comment