Member Discussion

Vector Based IP Camera Instead Of A Pixel Based IP Camera

We have been talking about the new HD-CVI versus IP Cameras and future max limits. One thing that we do not take into consideration (In my opinion) is that IP cameras do not need to be pixel based. Everybody assumes IP cameras mean pixel based. However, if you think about it, IP is just a communications protocol in the internet protocol suite. (Basically how it travels). I work daily in Adobe Photoshop (Pixel Based) and Adobe Illustrator (Vector Based). Both have their advantages. However, as days pass by, I find myself using Illustrator more than Photoshop (Not the case in the past). Why? Because my work in Illustrator is based on mathematical expressions to represent images. This means if I am working on a logo, my quality does not degrade when I zoom in, unlike a pixel based image. The size of a vector image will always remain the same (infinitely scalable).

After thinking about this, I decided to google this to see if others have thought about this. I found 1 university that is researching this (University of Bath) and applying it. Link - Example video from link (Keep in mind you're viewing from Pixel Based Monitor)

A new vector-based video codec could signal the end of the pixel.

Link to article -

Researchers launching a new vector-based video codec are claiming their work will lead to the death of the pixel within the next five years.

Professor Phil Willis, from the University’s Department of Computer Science, said: “This is a significant breakthrough which will revolutionise the way visual media is produced.

“However, to accelerate this project we’ll need companies from around the world to get involved. At the moment we’re focusing on applications in post-production and we’re working directly with leading companies in this area, however there are clear applications in web, tablets and mobile which we haven’t explored in detail yet.

“Involvement from a greater variety of companies with different interests will extend the project in a variety of ways and increase the potential applications of this game-changing research.”

My thoughts?

Can vector cameras become legit... Even if it is just a raw "vectory type" image... could it still pull in details from far away, like license plates (Remember Vector can be resized) or even the shape of a vehicle that can barely be seen on a pixel camera because it is blurry?

Can it be used for analytical purposes if it is not already used? (I'm assuming analytics currently uses pixels based images and using paths.)

Currently it only looks like vector can only be used after a picture is rasterized (Pixel). Can they make an image sensor that creates vectors instead of pixels?

Any Engineers that can expand, discredit, or fix my description or Bath University's research?

Ok, Jeremiah this may sound way out there, but the essential limitation I see in vector graphics see is one of direction. Every thing works great when creating ex nihilo. Throw a couple arc segments, snip 'em, join'em, with your basic Bezier thingy, then shading, gradients anything abstract thing you want until you have something pretty cool. Like a logo, or blueprints, 3d part explosion, etc. Then print it on a postscript printer and marvel how small the print file is...

But to go the other way, from having a cool rasterized image in your hand and wanting to just edit a circular primitive using your magic pen tool, is harder than pushing sales backwards thru the channel :) It's kinda like splattering eggs on a wall, sure you could recreate the egg if you had to, since we didn't lose any 'stuff', but how tuff is that? So vector works good for abstractions but try abstracting a human face into pure lines and shapes. And watch that little file go from 5% of jpg size (say for a logo) to 5000%... And still not quite capture it.

I think Godel or one his buddies proved that to map any reasonably complex terrain, you have to make the map as complex (big) as the terrain itself. And nature has noise and randomness everywhere you turn, so sometimes the best way to describe reality is just to "grin and bear it" and say its "a yellow pixel here, followed by two white ones, then one yellow-white one", as opposed to trying to abstract into geometric primitives egg splatter.

In your example see how you easily scale glyphs, but notice how abstractable they are, try doing that with a face... I'm reading the Bathian appeal for help now, but maybe you could say if and how you think they will overcome these obstacles.


You win because I refuse to engage in an ad stick hominem...

Your scaling is a off. When it comes through, you could have it come in at the smallest figure and the math is simply done locally. The bandwidth savings would be humongous, however, you would need more processing power on the client. It would also use less RAM than Pixel. If you can see a nice image in Photoshop, you can draw it in Illustrator. The color mixing, like skin tone is the tough part with Vector, however, Bath University believes they have or will crack the code.

As far as pixels. You state a white one and a few yellow ones. What happens when you zoom in to the pixels. Is it a quality pixel? How far can you zoom in? Or is it a blurry mess?

Vector covers every shape, line, letter and number. If you zoom all the way into an image, what is the shape of a pixel? Something that does not make straight line or a smooth curve...little squares? Something 8 bit?

What will probably happen is more of a hybrid type of video. Vector then being rasterized into pixels on the client end.

Trying to compute a vector image of a 2MP camera 15 times a second would probably take some pretty decent in-camera processing. Vector images also aren't the best choice when there are lots of subtle tone changes throughout the image, which is more common in a CCTV application.