Pointclouds : Boon or Bust?

by on Mar.19, 2010, under computer graphics, games industry

Therein Lies the Problem

This is a video of “Unlimited Detail”, a new technology from the company of the same name. They introduce their way of representing (and displaying) 3D data in realtime. Here’s the video :

My comments to follow…

This demonstration was interesting, once you dismiss the overtones of marketing and salesmanship. A search-based representation of pixels is a new concept (at least to me) that holds some amount of promise in the quest for a more efficient graphics system. To me, most games do depth very, very well. From Nvidia latest GPU pump to Carmack’s megatexture technology, there seems a clear and well understood path towards more photorealism in realtime. It’s only getting better and better as we progress, and even Carmack has stated that they’ve essentially “licked it”. To paraphrase his Quakecon talk of 2 years ago, he admitted something like, “we’re getting so good at this, it doesn’t seem to be much of a problem anymore”. I’m tempted to think that it’s not that interesting an issue to him any more!

To me, that really isn’t the problem anyway. The real issue in photorealistic gaming technology is not our ability to push geometry up on the screen. It’s our ability to create that geometry.

Games artistry is the most difficult aspect of any game that’s made today. The lion’s share of time invested in the production of games is well and truly in the balliwick of the artists, the poor beleaguered lot that patiently construct every last detail in the worlds you explore. And it’s only getting worse, not better. I know that companies employ all sorts of trick technology to speed up development time, yet despite very clever use of algorithmic production techniques (ie. SpeedTree, etc.) I do not see anything that promises to ease the burden on artists to any great degree. Something’s gotta give!

Handcrafted Worlds Must Die

And that’s where new technologies like Infinite detail’s “point cloud” system come in. The production of gaming worlds and assets goes back into the hands of mathematics, where I believe that it should be. This is not to insult the amazing talents of 3D artists at all – it’s just simply a matter of scope. As we “scale up” our gameworlds into larger and larger environs, filled with interesting and interactive “things”, it should be obvious how intractable the issue really is.

I wrote a long time ago (“The Next NextGen”, Gamers Quarter Issue#7), that perhaps one day we could “grow” our games. By that I meant algorithmically, much like the real world produces um, the real world! Based in a repeatable permutative system, a large set of parametric algorithms of sufficient sophistication will one day populate our virtual spaces, our games, our simulations. To me, Infinte Detail, with a little love and funding, seems at first blush like a healthy step in the right direction. It’s got a ways to go, but at least someone is doing something about it.

Comments are closed.

Looking for something?

Use the form below to search the site:

Still not finding what you're looking for? Drop a comment on a post or contact us so we can take care of it!