#925 overall
Computational Photography
YouTube
1 recommender · 1 thread · recommended in 1 year, 2020 · score 0.7
A YouTube lecture series by Marc Levoy, a Google computational photography researcher, explaining the theory behind techniques that combine many captured frames into one image, such as those used in Pixel phone cameras.
One commenter points to it as a great series of lectures on the theory behind Pixel's computational photography.
Science & Engineeringintermediatefreevideo lectures#computational photography#levoy#image processing#hdr#camera pipelines
Recommendations by year
20062026
The unholy quality on pixel phones is achieved through taking many pictures and combining them in smart ways (Google's computational photography wizard Marc Levoy has given a great series of lectures about the theory behind it: https://www.youtube.com/playlist?list=PL7ddpXYvFXspUN0N-gObF...) through a Qualcomm Hexagon processor, which has a special VLIW (Very Long Instruction Words) architecture. This co-processor is in every high-end Snapdragon (800 series) phone made in the last few years and you can relatively easily get the software to run on them through modded Google Camera versions commonly known as "Gcam ports".
My actual point is: You don't need special camera firmware to do this.
What you do need special (or open enough) firmware for is stuff like long exposures - the inability of which highly frustrated me on the phones, especially from Sony, I have owned so far. Doing weird stuff with the camera will probably be quite fun, but then again I don't believe the actual camera device will be that good.