Vector-based video could mean no more pixels
A newly-developed codec could make pixels obsolete within five years, according to its creators
Unlike traditional bitmap graphics, which are made up of an array of pixels, vector graphics consist of lines, curves and shapes that are based on geometric formulas. Not only do they take up far less memory than bitmaps, but sections of them can also be enlarged without any loss of resolution. Currently, however, vector graphics aren’t well-suited to photorealistic applications, such as video. That may be about to change, though, as researchers from the UK’s University of Bath have developed a new program that is said to overcome such limitations – the scientists believe that the technology could make pixels obsolete within five years.
The main problem with vector graphics is that they tend to be made up of sharply-defined areas of solid color, lacking the subtle transitions between those areas that are seen in bitmaps. As a result, the graphics are good for things like posters and animation, but tend to look a little cartoon-like.
The new codec (a program that encodes or decodes a digital video stream) is reportedly capable of filling in the boundaries between the elements in vector images. No details have been released regarding how the process works. The result, however, is moving vector-based video that is said to be equal in quality to bitmap video.
“This is a significant breakthrough which will revolutionize the way visual media is produced,” said Prof. Phil Willis, of the university’s Department of Computer Science.
The codec was developed in partnership with tech firms Root6 Technology, Smoke & Mirrors and Ovation Data Services. Commercial partners are now being sought to develop the technology further.
Samples of the vector-based video can be seen at the UBath link below.
Source: University of Bath via PopSci
About the Author
An experienced freelance writer, videographer and television producer, Ben's interest in all forms of innovation is particularly fanatical when it comes to human-powered transportation, film-making gear, environmentally-friendly technologies and anything that's designed to go underwater. He lives in Edmonton, Alberta, where he spends a lot of time going over the handlebars of his mountain bike, hanging out in off-leash parks, and wishing the Pacific Ocean wasn't so far away.
All articles by Ben Coxworth
The main thing is to ensure that this codec is open-source. That way, there will be no vendor lock-in.
This is a very big deal.
Transformative, in fact. The effect on gaming in particular will be stunning. I hope they can get it to 100% quality, the demo video was obviously hiding a lot.
But a great idea nonetheless, and certainly the market need is there.
I'm very skeptical. Following the links to the video showed nothing convincing.
Lets see a side by side of a large photo in both bitmap and their new vector format.
I can't see how a vector based codec can be more accurate at representing an image on display devices that still uses pixels. Bitmaps of sufficient resolution include unique info for each and every pixel on the display device.
Light field vector maps are a great advancement in imaging....
BUT... this may be a little sensationalised.
Pixels are the byproduct of the digital nature of photography (and video)
There still must be a finite spatial sampling step, and this means that the effective pixel limit will still remain....
Between the sampled variables, obviously it is possible to average the light vector field to obtain a "lossless" picture, however if the information which was not captured in that space was information, other than the average of the adjacent pixels, information will still be lost in the sampling method.
This technique is also done in digital filtering of images with traditional pixels, it is possible to re interpolate the pixels to increase the pixel density and then perform a gaussian filter operation (or other spatial averaging (or whatever operation is desired) and "recover" the lost data.
This being as it is, I will not be unhappy with increased image detail with a smaller storage size....
If I am very wrong, please, someone who knows the topic set me straight. (Not asking for a bunch of hate for being a bit pedantic.)
The ability to change focus points, or to focus on everything in the image, have multiple discrete focus points, or viewer selected real-time enhancement will be awesome. (If it is possible with the "limited" computing power now available...)
"...the scientists believe that the technology could make pixels obsolete within five years."
...until the image is viewed on a digital device.... Then PIXELS!!!
You know how nice that would be, if you are the family "photoshop guy" and get a 800x600 photo they want enlarged, and for the 4,394 time you tell them you can't blow it up too much without a HUGE loss of quality?
Rusty I feel your pain buddy... since 1997 at least.
Its back to the future! It will be like sitting at a colorized IBM 2250 from the 1970's.
Every graphic artist should post this by their desk. http://www.greystoneinn.net/comics/20040330.gif
@Rusty and Todd
"Can you blow this up?"
"Sure. Got any firecrackers?"
C. Walker Walker
co-processors for the math necessary (instead of memory) to process 4D chaos theory images? sounds like one of the first PC-XT maneuvers with hardware. deja-voodoo....
No this could still be useful, even and especially considering the final Conversion to pixels. Now when making a film everything must be done with an eye on the final display resolution. To little captured, your detail will always suck at higher res, too much and file size becomes impractical.
At higher res vector can be more efficient, since a lot of data is duplicated. Most video coded compressions are just kind of hacks that simulate vector representation anyway.
If you have a movie represented in vector formats though you don't have to transcode it again for each display device, all the math would be inherent education the codex.
"sections of [vector graphics] can be enlarged without any real loss of resolution." This is true...but it is also true of bitmaps.
The reason for this is that CCD's sample an image, and record information about the image, pixel by pixel - just as a bitmap does. So a bitmap, or the raw output of a CCD are both equally good inputs to software that converts to a vector image format. Once converted, the vectored image can be enlarged or shrunk and resaved as a bitmap. In fact this is essentially what GIMP does when you resize using the SINC (rather than bi-cubic) method.
I guess, lightfield-cameras are a much more convincing approach to more picture control. Adobe is working on one...
RE :"could make pixels obsolete within five years"
Does this mean 5 more years of research to get memory use & processing speed competitive with pixel based storage & presentation? Or does it mean nobody will be looking a pixels in five years.? Vector would probably be way more compact for a true 3D (like a hologram , not stereo vision) system than pixels, if such a thing could ever be done with pixels.
We live in hope. My complaint about my flight sim is that, as the article says, it creates too much contrast between things - because it is literally rendering those things separately. Even with awesome new anti aliasing etc, the image doesn't look like reality unless you squint your eyes.
I've just watched the video. For anyone who hasn't done so, I urge them to try the android and iOS app "paper camera". I think there is a free trial. It's pretty much my favourite app because it takes the camera feed and filters it in a way that reminds me of this article and the video on the university site. The video looks a lot like a paper camera filter applied to a normal video file. Like I said, we live in hope, but it seems a long way off to me.
The suggestion of unlimited resolution is false. While each vector element is 'resolution independent' this does not mean the data which the vector represents is. Regardless of if a video if represented as a vector or as a pixel, there will always be resolution limitations on the source side as no camera can directly record light as a vector. Even if this could be solved, optical resolution of the lens will always limit any imaging system.
Over 160,000 people receive our email newsletter
See the stories that matter in your inbox every morning