View Full Version : Vector graphic and video codecs

12-13-2012, 10:26 AM
Just came across this article, think some of you might find it interesting:

While the article talk about conversion from pixel to vector conversion, this will not increase resolution of course, just getting rid of pixel blocks on zooming.

It would be quite interesting to have a camera capable of recording directly to a vector codec, but I presume it would need a complete redesign down to how the sensor captures the images and significantly different internal processing, but maybe some of you guys with more experience with camera engineering would have something interesting to contribute with on this.

It is still an interesting thought, with all the discussions on 2K, 4K and higher resolution video going on various forums.

12-13-2012, 11:34 AM
That would be a CPU hog no doubt!

12-13-2012, 12:03 PM
The whole point of the BMC is that it gives us access to RAW sensor data. Why would you want to store something that has been derived from pixels into something that is defined by vectors? Doesn't make sense to me. Wavelet transforms make far more sense when coming from a pixel based source (redcode / cineform / jpeg2000). Motion graphics maybe.

Pretty sure h.264 uses motion vectors to separate background from foreground objects within a scene to increase compression ratio without loss of important detail. But that's temporal vectors rather than instantaneous spatial ones.

Thanks for posting, really interesting stuff.

- Darren

12-13-2012, 12:42 PM
Absolutely, I don't see any need for this for current cameras or sensors.

What I was thinking about was for radically different cameras and future sensors that could potentially take advantage of a different way of recording data. If we were to talk about just converting from a traditional pixel based source like you mentioned, the advantages would not go far beyond what was already outlined in the article.