From: Alan Taylor (yoame_at_WOLFENET.com)
Date: 12 February 1999
Warning - this is very long and self-indulgent. I've been quiet lately, but haven't stopped thinking about things. I'm posting here and am wondering whether this might be a useful contribution to the www-vrml list as well. Please let me know what you think.
Okay, my problems with VRML _aren't_ really anything that can be easily overcome by a better language, a simpler browser plugin, or an easy-to-use creation tool. My problems are the same ones we all know and acknowledge and run into every time we show our work to someone not versed in VRML -
I'll address these one at a time (this is all purely my own perspective and opinion).
That's a fact - so is much of the web - people are learning to get used to it from a download standpoint, but not from a display standpoint. If you show an average user an interesting VRML environment right after they've been running around in a Quake world or some other wonderful game, there is no competition.
I don't exactly know the technical reasons why VRML is so much slower to render realtime than proprietary formats in games, but you know what - it really doesn't matter. I can list off a dozen facts that explain why VRML is slower because it doesn't use a binary file, or because it's run in a plugin, or because it's a poorly-authored world, but to someone who is dropping by for a visit, those are only poor rationalizations that really don't mean a thing.
The experience is what it is - slow and poor in comparison to other similar-looking technologies. This is not griping - I am very used to the differences and work hard to accomodate them in the worlds I've built, but it still doesn't change the fact that my best friend (a non-VRML guy) thinks the VRML experience stinks.
2. "It runs unpredictably"
I should retitle this one "It fails unpredictably", but I was being charitable. My experience with the few worlds I've created has been consistent. I work and work and work on one machine (at home). When I reach a stage where I'd like to test, the world looks different on another machine, let alone the difficulties between the many browser-browser combinations. If a world I work on gets beyond a low-level set of interactivity, the chances of cross-platform failures go up exponentially. Even when you've written up the most wonderful cross-platform, well-tested smooth-running world, you still have incompatibilities.
When I finished and launched the International Space Station model on MSNBC, I had the opportunity to view that model on well over 100 machines - PC, Mac, an O2 (don't know what flavor Unix it uses), using both CosmoPlayer, WorldView, and the MS VRML2 viewer. At least 8 of those machines had terrible problems, even with the correct software. There were a couple of video driver bugs, a couple of low-powered machines that just seized when presented with the model, and one that inexplicably blue-screened and died whenever it hit a web page with VRML on it.
Of those machines that did have things eventually work, many of them had old VRML browsers of various flavors (even a few "media players" that claimed to understand VRML, but didn't). More than half did not automatically understand the VRML file (didn't have a pre-loaded VRML browser). A couple had installation problems, which I think were related to the individual machines, not the plugins. But the situation of the painful download still remained and I installed many a browser over those weeks so friends and colleagues could see what we had out there.
I've mentioned this before on the www-VRML list, but it was made clear to me by my management that the ISS VRML model was an experiment and that it was viewed as a liability, if not a partial failure. The logfiles I've seen tell me that anywhere from 1/2 to 1/3 of the users who hit our ISS intro page go in and actually get the model file (one more click in). That tells me that the "You must have plugin 'X'" message and the "If you see the spinning cube, you're ready to go" test on the intro page scare off most of the people who were curious enough to go there in the first place.
If this were work for a paying client, and not my permanent employer, I'd likely be doing a lot of damage control - at least I'd have a dissatisfied customer, and would shy away from ever suggesting VRML as a solution again. As it stands, I am still pushing (though quietly) to continue VRML work here at MSNBC, along with other multimedia - but this is all tempered with a growing internal frustration with using plugins in general (Windows Media Player, Flash, and VRML).
2. "It's clunky"
People (me included) are spoiled by "A Bug's Life", "Jurassic Park", etc. Even realtime 3D videogames make up for large low-polygon models by using rich lifelike imagemaps. Every time a creator tries to ramp up creativity, richness, or visual beauty in a VRML world he/she is punished in a rather harsh way. I understand that this is a limitation of creating _any_ 3D virtual experience, but it always seems to me that VRML gets hit worse than any other.
I once tried out a small (8k) background image for the sky/ground for the first of my RavenStories. The difference in performance between the world with or without that image was phenomenal. The fps went down by half or more. Even smaller images just ground the fluidity and illusion of animation down to a level where I just swore off texture maps in my second RavenStory. The result was a cartoony look that worked well enough, but wasn't my first choice, and wouldn't be my best choice. I have since been delving into 3DStudioMax, Bryce3D and Rhino3D to better express what I was after - you can peek at a test image here ( http://www.geocities.com/SiliconValley/Orchard/3767/akscene1.jpg ) if you like.
Yes these solutions aren't interactive by nature like VRML is, but I would feel safer (more assured that I could reach a broad audience with a minimum of trouble) making beautiful still images and scripting interactivity like Myst or Riven than I would making a VRML world at this stage - for many of the reasons listed above.
I'm not abandoning VRML as a medium, but its limitations have started to drag down my interest and have begun to impact what I'd like to create. Cartoony avatars in primary-colored boxes won't ever be as appealing as texturemapped environments and beings. (I know that's hyperbole, but I'm making a point).
I really don't think that at this time the advantages of Real-Time 3D interactivity outweigh the advantages of beautiful pre-rendered 3D or the simplicity and effeciency of 2D interactivity like chat rooms. That's not to say it never will, but not now, IMHO. This points at the base conclusion that I've seen brought up on the www-vrml list, that there aren't compelling business reasons to adopt this now, so nobody is picking it up. For me, a "compelling business reason" doesn't even have to be about making a monetary profit, and has more to do with "is this a waste of energy", or "can I do this in a more efficient manner".
4. "It has very little content out there that can be pointed to as useful or compelling"
I would never want to disparage other's work, especially since I understand the efforts that go into it, so I'll look at mine. The ISS VRML model - I don't really know how much more efficient or useful it is as a teaching tool than would be a series of well-defined images in a news magazine. A magazine certainly is more accessible and less prone to malfunction. How compelling is the model itself - Frankly, I think it's pretty good, for a VRML model. But compare it with some of the pre-rendered work that NASA has distributed to the worldwide press and it really pales in comparison. The only distinct advantages I can see are the ability to interact (spin it, move around, etc.) and the "feel" of seeing it with stereoscopic effects. But at what cost - a percentage of a percentage of a percentage of the population even _can_ see this, fewer still would want to see it, and even fewer have seen it. Those that do have to wrestle with plugins, 3D interfaces, etc.
Ugh. It makes that magazine spread look pretty good from the standpoint of "effort made" balanced against "information conveyed".
Ravenstories - I love 'em, and had great fun making them, but I can't think of a single reason why these wouldn't be better looking, or more compelling as pre-rendered movies.
When I read the www-vrml list, I get hopelessly lost in the technical jargon, the difficulties of the Java EAI (which I've never embraced - I think it's wrong headed, but that's a different rant). To be frank, I don't know what most of the talk is about. I don't fully grasp the VRML-UMEL, or VRML-NG, and especially have trouble grokking X3D... sigh. This is due more to my ignorance and non-technical background than to any failing I can see on the part of those championing the above acronyms.
I'm not unintelligent, I'm just not that interested. When I started making VRML worlds, I didn't do so to join a clique, or to become a member of a society, or to join in a grass-roots effort to define a language and its deployment. I did so because I thought it was cool. I made my first worlds because they were fun and the gratification was tremendous.
I had a great chat with Mark Pesce at Siggraph98, and felt good that I was able to thank him for championing this technology and inspiring me on several occasions. I felt flattered that he told me that the RavenStories had given him some inspiration as well. That's what it's about for me - creative people bouncing ideas and stories off each other and sharing them with others. I have no interest in defining a language spec and getting involved in political struggles.
Sometimes I feel like an artist in a room full of bureaucrats, all arguing about the makeup of the canvas, how long a brush should be, what to name the pigment colors, etc. - when all I want to do is paint. I really do not want to attack or disparage those involved in the efforts to make things better, but have to admit that the discussions are often either over my head or outside my scope of interest.
I'm not walking away from VRML, but think you could say I'm "on hiatus" for a while. You know what I really miss - the contests like Cosmo's BUZZ WRL of the week. Seeing good stuff every week, spurring the competitive juices, and getting all sorts of ideas in that manner was more inspiring for me than any other event before or since.
For those of you reading this far - thanks for bearing with this rant, I'd
love to hear from you too,
-Alan Taylor
This archive was generated by hypermail pre-2.1.8 : 28 November 2004 CST
