Hugo icon Linode icon

Microsoft Pix

Sep 30, 2018
2 minutes read

Note: Most images on this page have been picked from the official landing page.

Microsoft Pix is Microsoft’s super camera app that rivals those of the iPhone and Pixel devices. Developing this iPhone app was a very enriching experience for me where I learned aspects about product development, merging cutting-edge technology with user experience and balancing code quality with pace of execution.

A screenshot of Microsoft Pix from the product landing page.

Picking the best shot

A few components working in tandem made this possible: combining a stream of photos, focussing on people for sharp images and adjusting settings when a face is detected.

Picking the best image


Microsoft’s Photosynth was a really cool demo back in 2008-2009 during my undergrad. It now lives in the app’s Panorama mode and works in near realtime!

Microsoft’s Photosynth

Live images

Subtle motion in images. This was my hack at Angelhack in 2013 and here I am working on a product that already does this. Such wow much fun! Live images in Microsoft Pix


We hacked something at Microsoft’s One Week Hackathon and created a feature worthy of shipping. This was probably the most fun thing I’ve ever worked on. Pix-Comix

Whiteboard and business card recognition

Pix can detect, capture and clean-up images of whiteboards and business cards. The app also features directly connecting with users on LinkedIn based on their business cards. Very slick! Pix whiteboard detection


Speeding up videos without giving you vertigo was the plan behind this project. Stabilizing videos on device and then encoding it to a video was a feat accomplished by this project. Pix Hyperlapse

Style transfer

Style transfer is a very popular machine learning technique that makes an image look like a work of art instantly. As an addon, Pix Painting lets you create artistic “painting in progress” videos.


  • Product development is hard.
  • Writing cross-platform + performant code is hard.
  • C++ is a lot more fun that it used to be a decade or so ago.
  • Machine learning + computation photography can produce awesome results.

Back to projects