Comments on: How Far Away Are We From Vector-based Photography?

By: MixedMoj

MixedMoj — Tue, 11 Aug 2009 16:36:13 +0000

Great idea, but to be honest with you, the amount of processing power required to produce a vector based image output would be tremendous, compared to a raster image. Approaching “artificial” intelligence, even. It takes a line artist hours to render even a moderately detailed vector, imagine having to duplicate the level of detail required for the average 12 megapixel image. There’s simply too much fine detail. Although, I do agree that vector-based printing technologies would be the next logical step.
Don’t get me wrong, I love the idea of vexeling, or rather, the marriage of both pixel and vector based rendering. But the bottom line is, fine detail is what makes raster based graphics the better choice for digital photography. Photography and line art: two completely different things.

By: 11 Little Cowboys and Indians

11 Little Cowboys and Indians — Tue, 11 Aug 2009 16:36:13 +0000

Yes vector files are scalable but we would still need to think in terms of “resolution” in the sense that the files would only be good as the level of vector detailing put into them.
Think about converting a 50×50 pixel file to vectors where every pixel or gradation in color has its own vector. It would be scalable but of very limited detail. Doing this on a larger file, lets say 2000×3000 pixels would result in a staggering file size – scalable yes but again the detail would be limited by the number of vectors used to draw or map out the image. Photographers and printers would still always be striving for more detail as we are today.
Obviously raster files are the most efficient way of doing this so I think they will be here for quite some time to come.

By: Merlin

Merlin — Tue, 11 Aug 2009 16:36:12 +0000

A picture can’t really be described in vector graphics. Vector graphics would describe an image in terms of lines starting at X Y coordinates, going a certain angle, for a certain distance. Further description could include width, or color. It’s display on the screen is scalable because it’s appearance can be calculated based on the definitions of the object. This is very efficient for line based drawings because a few bytes can be used to describe an object that covers a considerable distance. The object being described is not real though. It is a digital description of a theoretical object.
You can’t do the same with a picture because your original source data, the subject you are photographing, is analog, not digital. There are continuous changes in the luminance and color values from one pixel to the next. Even if you shot a plain white wall, you will discover that there are shading changes across it, and small variances that give the image texture. If you removed them with your paint program you will have noticeable blemish in the image that looks very unreal. If you put something as complex as a person in the image the changes will be everywhere from black to white and all over the color spectrum. To try and reduce that to a vector description would create a cartoon look rather than a photograph. To preserve your detail the vector description would have to describe each pixel individually, but then it would not really be vector graphics anymore, it would be bitmap and nothing would be gained.
It is not a question of the speed of processors, it is driven by the nature of the data you are trying to display. The reason you can’t scale the picture up without loss of quality is that you are asking the computer to display data that was not originally captured by the camera. Current imaging elements are very good and in many cases exceed the resolution of the lens on the camera. You can’t ask the computer to display data that never made it through the lens. You can only ask it to calculate a probable value for what was between the pixels it did capture.