By applying ideas from level set methods, we can represent topologically changing scenes with NeRFs. HyperNeRF handles topological variations by modeling a family of shapes in a higher-dimensional space, thereby producing more realistic renderings and more accurate geometric reconstructions. This paper was published at SIGGRAPH Asia 2021.

We present the first method capable of photorealistically reconstructing a non-rigidly deforming scene using photos/videos captured casually from mobile phones. Nerfies turns selfie videos from your phone into free-viewpoint portraits. This project was published at ICCV 2021 and was covered by several online magazines: Gizmodo, Two minute papers, Wired

Our paper describing the core technology behind automatically picking the best quality image from a burst of captures was accepted at IEEE Transactions on Image Processing. This is used in the Microsft Pix app.

Our patent describing a technique to create a summary seqeuence of images from a video was approved! This technique was used in Microsoft Pix’s comic strip feature and is the result of us winning an internal hackathon at Microsoft!

Our book titled OpenCV 3 Blueprints is now public. I contributed a book chapter on video image stabilization using computer vision signals in the form of feature matches across frames, using motion estimation using the camera’s built-in gyroscope and rolling shutter estimation.

My last movie working as a Technical Director in the surfacing department at DreamWorks Animation!

The Penguins of Madagascar return as I worked in the surfacing department!

The second How to train your Dragon was released where I worked as a Technical Director in the surfacing department!

My second Hollywood movie credit working as a Technical director in the surfacing department.

My very first Hollywood movie credit! I worked as a Technical Director in the surfacing department.