Disentanglement Is the Next Deepfake Revolution
PhilNote: easy-read long article with many examples.
New research from Israel proposes a novel method of using synthetic data, such as CGI heads, to bring deepfaking into the 2020s, by truly separating facial identities (i.e. the essential facial characteristics of ‘Tom Cruise’, from all angles) from their context (i.e. looking up, looking sideways, scowling, scowling in the dark, brows furrowed, eyes closed, etc.).
...
this represents a way forward for a fundamental separation of instrumentality (such as ‘change the angle of the head’, ‘create a frown’) from identity, offering a path to a high-level rather than ‘derivative’ image synthesis-based deepfake framework.
The new paper is titled Delta-GAN-Encoder: Encoding Semantic Changes for Explicit Image Editing, using Few Synthetic Samples, and comes from researchers at Technion – Israel Institute of Technology.
See the full story here: https://www.unite.ai/disentanglement-is-the-next-deepfake-revolution/
