Multi-Faceted Sentry Bot

SentryBotScreenShot

Having succeeded in my attempts to capture data from Doom3 (see comments in earlier post – Link), I decided to start playing with the captured .obj file… mostly to see how easy (or hard) it would be to get the geometry to go “solid” in CAD. For me the most intriguing object in the entire game was the little critter above, a Sentry Bot, so I decided to pluck one out for some experimentation.

Below are some screens to show what the mess looked like after importing the file into Maya.

SentryBotMesh-a

The wide angle view of everything captured is shown in the view above. The model’s spherical rotation handles make a nice locator here.

SentryBotMesh-b

Zooming in on the model you can start to see the walls of the room in which the character is positioned.

SentryBotMesh-c

Within some odd “blur-triangles” that permeate the mesh I found the Sentry Bot.

SentryBotMesh02-FootScaleweb

And here’s the little guy extracted and scaled (each grid is a square foot). Click on the above layout to see a larger view.

After I got that far I started playing with converting the mesh, but I’ve not been happy with the results so far (and it takes over an hour for a Sub-D to NURBs conversion). I have something I want to try, so I’ll be getting back to this soon. Also note that while I could have gotten a geometry screengrab when the model was in a symmetrical pose, I thought something dynamic would be more interesting. We’ll see how that turns out (it’ll most likely make the CAD conversion more difficult if what I’m planning to try next doesn’t pan out).

{Image source: Doom3 © id software}

3 thoughts on “Multi-Faceted Sentry Bot

  1. great work! any sense of what all those odd “blur-trinagles” might be? I have only seen them in Quake4/Doom3 — do you think they are real triangles from the games, or could OGLE be mis-capturing some data?

  2. Yeah. I have a possible idea about what might be going on. I’m going to try testing a theory today (hopefully). I’ll comment on what I find.

  3. Suspicians confirmed and I think I might have an answer.

    When I grabbed the above data, I max’d out the graphics settings for my machine. When I drop them down to minimum, I get a nice clean mesh without all this extra stuff.

    What I think we have here is a dynamic occlusion mesh of some kind tied to the graphics system (at least); probably a part of the dynamic lighting and/or normalmap vector displacement calculations.

    Very, very cool.

Comments are closed.