Bandwidth

It used to be alleged that the number of cars will always grow to fill the space available, thus reducing the incentive to meet the demand for roads but providing an incentive to charge for roads. If motorists had to pay for the road as well as the petrol they might calm down a bit. On the other hand, the costs of charging might be rather a lot, so perhaps we had better wait until the (technology rich) congestion charging scheme for London makes a profit before making a move.

But today's thoughts are not about the capacity of our roads, rather about the capacity of our broadband infrastructure. In fact, about a wheeze to burn up some of that infrastructure.

As things stand we all share a lot of pictures. Our happy snaps, often at megabytes a time, are copied and sent all over the place. But happy snaps are becoming boring and this punter wants more than just a two dimensional rendering of things. I want a three dimensional rendering, coupled with a viewer which can change the position and direction of view, a rather snappier version of what I sometimes get with Google Streetview (my experience of this product being very mixed. Haven't got the hang of it at all, partly because I can no longer reliably find where it lives).

For phase 1 of this project, we suppose that whatever it is that I want to see can be expressed in terms of a finite number of objects populating a viewing space. Leaving aside boundary issues, we suppose that we are only interested in the exterior of these objects and that these exteriors can always be mapped onto the surface of a sphere and so triangulated. That is to say that the surface can be agreeably expressed as a network of small triangles and that our new generation image stores the position and colour of each of those triangles (of which a good quality image might need millions). We then attach a bit of code to that image which constructs a flat, two dimensional image corresponding to any nominated point of view and then displays it on the screen in the ordinary way. We leave aside the interesting question of how the code knows which side of the triangle is the outside, but I imagine that most of the machinery to do all this must already exist in the context of computer games (about which, I should add, I know more or less nothing. Don't know how to play them, never mind know how they are put together). Add a joystick or some such contraption to my PC and I will be away, able to roam around my new generation image at will.

There might be efficiency gains in, rather than storing just the colour of a triangle, storing an image. So something like a patterned carpet might be stored as just a couple of triangles (need at least two to do a rectangle), rather than having to break that pattern down into triangles of a single colour.

We leave it as an exercise to the reader to work out how one generates a new generation image, perhaps starting from a small number of old generation images, taken from enough different positions and angles to capture the objects in question.

We leave aside the need for an integrity checker, something that checks that all the triangles in my new image are coherent, that they join up into something sensible.

We should probably start with a world made up of brightly coloured building bricks, the sort of thing that infants like. Having cracked that, we move onto steadily more complicated objects. A person would be reasonably testing, a tree perhaps even more so. Or perhaps a strand of DNA or a molecule of protein. No doubt, the manufacturers of pornography will follow developments with interest. The mind boggles.

For phase 2, we move onto moving images. Either images which move of themselves, as in a movie, or images which move on command, as in a computer game. Roll over please, I would now like to see the back view. But maybe that had better wait until we have got a firmer grip on phase 1.

In the meantime, I want to get onto the international standards committee which sets the standards. Must be plenty of foreign jollies to be had in that department. Not to mention all the interesting images which one would get to look at.