Researchers at the French research institute Bcom, with the aid of a wunderkind plucked from a nearby university, have developed software that converts existing SDR (standard dynamic range) video into HDR (high dynamic range) video.
That is, the software can take almost all of the colour video content produced by humanity over the last 80 years and widen its dynamic range, increasing the brightness, contrast ratio, and number of colours displayed on-screen.
I’ve seen the software in action and interrogated the algorithm, and I’m somewhat surprised to report how good the content looks with an expanded dynamic range.
But garbage in, garbage out, right? You can’t magically create more detail (or more colour data) in an image. Well, you can—Google produced detailed face images from pixelated source images—but philosophically it is no longer the same image. When a film is cropped for TV broadcast, or you receive a blocky low-bitrate stream from Netflix, or Flickr changes the JPEG profile on an uploaded photo… are those the same image as the artist/director/videographer intended? Or are they different?
Read 17 remaining paragraphs