You are here

Apparent shape manipulation @ Siggraph 2016

8 Dec 2016

One of the things we've considered in the PRISM network is whether we perceive the appearance of objects in images (their shape or material for instance,) directly from image features available to the visual system.If this were true, we could modify an image using some smart operators to mimic changes in shape or material, without having to rely on any additional 3D information.This is what we have demonstrated in a Siggaph paper in 2016: the apparent shape of objects in images can be modified by a warping algorithm that manipulates the cues humans are likely to rely on when perceiving shape (compressions and stretching). The method relies on a differential analysis of the image formation process that has been developped throughout the duration of the PRISM-ITN, and which connects the dots between light transport on the one side and visual perception on the other side.The full Siggraph talk and Q&A session is available on the ACM Digital Library, here. (access restricted to ACM members)