Vector Graphics are back baby!

[buzz_n64]

Brit researchers aim to replace the pixel Vectors are the way forward
http://news.techeye.net/science/brit...lace-the-pixel

"The age of the pixel is fast coming to an end as a team of British researchers replace the technology with vectors.

According to ExtremeTech, vector graphics have been around for a while. They are made out of geometric primitives, are infinitely scalable, making them the ideal image format for illustrations, clipart, maps, typography, and Flash.

However, pixel bitmaps have control over streaming videos, digital cameras, movie editing, and video game textures. This is despite the fact that as display resolution increases, so does the number of pixels and larger bitmaps are taxing the computer. They can’t scale and changing formats is rubbish.

Now Philip Willis and John Patterson of the University of Bath in England have come up with a codec that replaces pixel bitmaps with vectors. Dubbed Vectorised Streaming Video (VSV) it converts a bitmap image into vectors.

The pair had been working with vectorised photographic images converting bitmap images into perfect, vectorised copies but this gets things, literally, moving.

The Bath researchers are working with Root6 Technology which is an outfit that specialises in transcoding, as well as Smoke & Mirrors which is a post-processing studio, to bring the codec to market. Smoke & Mirrors claims there will be working demonstrations of VSV within the next three to six months. It should take five years for the pixel to go the way of the Norwegian Blue.

The algorithm is similar to the auto-vectorisation tools, such as Adobe Live Trace. Willis and Patterson claim to have fixed the problem of colouring of spaces between geometric shapes and made things photorealistic.

It could also mean that it is possible for cops to go to their computer screens and blow up crucial pictures, just like they have done in the movies since the 1960s."

What's old is new again it seems. It looks like advancing in gaming technology still has not reached its full potential.

[kedawa]

Rasterizing from a vector isn't really vector graphics. It still ends up as pixels.

[Oldskool]

I always liked Voxels. I wish they would have worked on that technology a bit more.

[Ed Oscuro]

Voxels are just a representation of a three-dimensional volume; they still need to be physically represented, either as pixels or some kind of geometric shape (like a line on a vector display).

With the progress in CPU performance seemingly slower than it used to be, it seems tempting to jump to something else, but I don't see how vector graphics could possibly replace bitmaps in certain uses. It would be nice to have the choice but I think that what it comes down to is that you'll need two displays or monitors, rather than one, if you want to use vectors for text and generated artwork. I don't see what's wrong with the current approach, translating vector graphics to bitmaps, other than the slight loss of quality due to grading. While working natively with vectors will allow you to save some bandwidth and processing on your device, it'd be pretty ugly to try to turn digital footage (I have trouble seeing how you would use vectors to represent a bitmap, like a video frame or phone camera photo) into vectors, and will probably end up needing just as much CPU time as before, and probably more to work on a vector display. One of the major wins of vector graphics has been in representing certain fancy shapes, like country flags on Wikipedia, with a small file, and that's already done pretty well.

So I think that going from a vector data structure to a bitmap display is a lot simpler than going from a bitmap data structure to vector hardware.

If you could swear off having modern conveniences like camera video phones, then I'm sure it would allow you to make a nice and easily readable device without little bumps...but we're already on the cusp of the "retina" display, and I don't think the average person is ready to deal with having pictures (not the hand-drawn kind) look clumsy on their devices.

All the stuff about blowing up things for the cops; now that's rubbish. There's no new data in the image. They can't do anything with a vectorized bitmap image that can't be done just as well or better with the right upscaling algorithms, unless they want to use it to emphasize lines and shapes, but surprise, the human brain is already pretty good at doing this, and interpolated (i.e. invented) data, such as the researchers are making, could provide apparent detail that proves false in critical situations. "Our great software has traced out the smoking gun!" "No, that's a teakettle..." It seems to me that a large part of what they're doing is just a riff on edge detection.

Anyway I better be quiet before they demo it, but seems totally premature and has dubious benefits.

[jb143]

I always liked Voxels. I wish they would have worked on that technology a bit more.

They've advanced to the point where they're not the recognizable Comanche style voxel graphics anymore. Unreal Engine 4 and all versions of Cryengine use voxel technology. Not to mention 3D Dot Game Heroes, Minecraft and all it's clones.