Last week we discussed briefly the production of still photographs in high dynamic range. Just recently, there has been some activity in doing ultra high fidelity video, using all the pieces of hi-def, plus increasing dynamic range and color bit depth. Whether these technologies will make it into our living rooms in the near future is anybody’s guess, but they probably will see some type of implementation in the long term.
The upside we’ve already covered a bit in last week’s post. Moving video to a dynamic range that more closely resembles what the eye is capable of perceiving is a good thing. In the case of TV, it would resemble more “looking out the window” rather than at a TV screen.
The short video sample I posted last week was produced literally as a series of still HDR photographs (processed as I described last week), resembling more time lapse photography (which it truly was) rather than actual video. If the same technique were to be used to produce video, the camera would need to take at least 3 pictures 24 times a second, or 72 frames per second to equal the current cinema standard of 24 fps. (See discussion on this frame rate in this previous post.) Then, that data stream would have to be processed in real time combining the over and under exposures of the same frame into one and sending it down the wire 24 times a second. That’s alot of processing horsepower, and because of the complexity of the process, I do not think that this technique will ever be used for serious video production.
There are, however, those out there who are experimenting with High Dynamic Range capable CMOS sensors and associated A/D codecs. There are some prototype sensors out there that are flirting with the 12 f-stop dynamic range specification, which is very close to the dynamic range of the eye. But even with this vastly simpler technology, because of the additional digital information required, changes will have to be made.
One such camera is in the hands of Professor Alan Chalmers of the University of Warrick, UK. With a 1080p resolution, his camera produces 24 meg of digital information per frame. By my calculations, that would mean that a minute of footage would require 36 gig of storage space, and the superfast network architecture to support that storage/transmission. This is 27 gig more per minute than standard uncompressed 1080p HD video. Transmission and storage over existing infrastructure would require compression algorithms of nearly 100:1. (See discussion of video codecs here.) That kind of compression, while theoretically possible, is certain to come with its own artifacts and compromises.
Production would also have to change. As we all know, some actors and actresses have not benefited from the increased resolution of HD video. This problem would be exacerbated in HDR. More care (read money) would also have to be exercised in production of HDR; the guy behind the Rockefeller Center studio window making hand signals will be seen.
I personally hope that some of these technologies do take hold. Increasing picture resolution through HD technology over the last few years has made a vast difference in the viewing experience for most of us. I suspect that these next iterations of ultra high fidelity video have the potential to be at least as significant as hi-def has been in the last decade. Present examples of this technology actually compress the tone-maps to make them viewable within present media constraints. When/if genuine HDR television comes of age, we will see that HD was just the beginning.
Until next time….
Loading ...