Slashdot Mirror
Review Of 3D Web Browsers
shelflife points to this very intersting article on the 3D Web browsers in Scientific American. He writes that of the 3D systems mentioned, "A Swiss company, Geonova (www.geonova.ch), seems to demonstrate best that the idea of a geography-based Web is feasible with today's PCs. Engineers there created two impressively detailed models of Switzerland--one of the entire nation with 25-meter resolution and another of two central cantons at 50-centimeter resolution. .. Text and iconic labels hovered quite legibly above towns, lakes, companies and tourist attractions; clicking on the labels opened associated Web pages. What other 3D browsers are there -- VRML plugins have been around a while -- yet they do not seem to be successful. Why is that?"
Why aren't VRML plugins used more?
1. It's slow.
2. No real or perceived value-add from using it. Why should I switch to a 3D filesystem browser, for example, when what I have works just fine?
Make it faster and give us a good reason to use it, and it might get more widespread use.
"This message is composed of 100% recycled electrons."
I'm not too impressed with anything I've seen from VRML, but Adobe's Atmoshpere was a bit of an eye opener. Check it out yourself at Adobe's Site ... free to play with while it's in beta. I haven't personally developed for this yet, but I hear it's quite intuitive. As a bonus, you have site-specific chat. Now if they will just add site-specific-VoIP I will start taking a more active interest, but for now I think all of this 3d Browser stuff is a bit academic as noone has found any particualarly good use for this stuff yet (at least nothing that the mainstream surfer is going to clamour for).
The demo is really neet - it's worth firing up a Windows box. As you fly around, the system feeds you better and better textures for the ground. Unfortunatly, it seems that it dumps the good textrues as soon as you fly away too far. The user interface works ok - user '-' and '+' on the numeric keypad to change elevations.
Moneyed corporations, non-working 'poor' and criminal prisoners are turning productive citizens into tax-slaves.
The XML solution on the horizon is X3D. This has a much better chance of being a useful 3D markup language because in most cases it would be only one representation of the base data among a set of alterantives including vector graphics (SVG) and XHTML.
As someone a little bit closer to this issue (check webpage) this is why I think VRML failed.
1. VRML-97 is not a superset of VRML-1. There are features in VRML-1 that don't convert easily to VRML-97 so people who started with VRML-1 had to re-do a lot of stuff by hand. That discouraged a lot of the early movers.
2. The VRML-97 specification specified too many things that didn't need to be specified (like text layout, which looks crappy in VRML anyway) and initially failed to specify some things very well. There was some question about what scripting should be used at first, later Java and ECMAScript worked their way in but that leads me to...
3. It duplicated things that could be done with other things. In particular, you can do a lot of 3d with Java, and if you are going to use Java to script your VRML world anyway you might as well just do everything in Java which leads me to...
4. Crappy installed base. Really weak VRML browser shipped with IE and Netscape died before its decision could have made any impact.
5. Somewhat different computing paradigm. The VRML file contains "sensors" which trigger events that are processed by scripts. In other words, the data drives the code instead of the code driving the data. Is it a file format? A programming language? What is it? I'll tell you, introducing a different way of computing is fine, but they didn't pitch it that way, which tells me that it was more of an accident. It's always a bad sign when different ways of doing things get introduced by accident.
6. Bloated syntax. I know I'll catch it from some people for this, but I stand by it. Why was the proposal for VRML-97 called VRML-2? I'll tell you: because it has twice as many brackets and braces as VRML-1, and it doesn't really make things any easier to read.
7. Performance, performance, performance. A few months ago someone on comp.lang.vrml posted something that looked like a simple Quake level. It ran at 1 FPS on my box in a tiny little window. The same box runs Quake full screen at least 24 FPS, probably more but I can't tell and I don't care because Quake looks fine. The VRML performance problem is intractable too, because it doesn't have any standard way to do BSP or any of the other tricks that games do.
There are probably other reasons too; that's just the top of my list. Oh well, I had a lot of fun with it in the early days, and I learned a lot coding for it but it is DEAD, DEAD, DEAD. I use my VRML program mostly to create animated GIFs and for photo-shop like effects (layering translucent PNGs and taking screen shots is cool) and I keep the web page up because I hate to kill stuff. I harbor no delusions. VRML will never see mass appeal. It seems to have carved out a niche in some government and academic circles, but there is no excitement there, no profit, and not much life.
For all intensive purposes, "whom" is no longer a word. That begs the question, "who cares"?