Why is Sports Photography Hard? (and what we can do about it)

Опубликовано 17 августа 2016, 1:07

Of all the genres of photography, sports and action is one of the most challenging. Especially for organized team sports played on fields and indoor arenas, the photographer is operating at the limits of pixel sensitivity, burst rate, focusing speed, and lens capabilities offered by modern digital cameras. Of every 1000 pictures the photographer takes, 10-30 are typically usable, depending on experience and luck. In this challenging problem the technologies and algorithms of computational photography and computer vision offer intriguing possibilities. I will start by describing more or less quantitatively what's hard about sports photography. I will then enumerate places I think clever and customized computing can make a difference, especially in focus and tracking, triggering of frames at 'decisive moments', and ameliorating the adverse effects of cluttered backgrounds and poor lighting. The talk will be illustrated mainly by my own bad photographs (and a few good ones) of Stanford sporting events. Time-permitting, I will also summarize my lab's work on programmable cameras and describe the Stanford Frankencamera and FCam API, but it is not the focus of this talk.