Digital Mercs

I’m not going to brag or anything but we’re basically kicking ass at this point. It took some colossal effort over the last few months to get this ball rolling, but now we’re well under way. Jon is getting tons of requests to put together a video tutorial on how we got our car working, and it looks like he’s gonna bust that out when he rigs up Johnny’s car.
On the topic of Johnny’s car, how great is that coming out? I was worried at first when he chose to make that as his first truly next-gen asset. I mean, I wussed out an made a broken piece of concrete… He’s stepped up and proved that he’s got what it takes to be a game artist in today’s crazy-fast-paced industry.
Anyways, enough gloating about my friends… After work today I sat down to crank out some textures for major track assets… I feel like I’m starting to really understand the power of the UDK material editor.

Throwing some assets into a scene...

I’ve spent a fair bit of time over the past week or so brushing up my UDK skills. I’ve scoured the internet for answers to all sorts of problems. If anyone out there is looking for a good introduction to Unreal, and its capabilities I can’t recommend the videos from Eat3d.com enough. They are heavily production focused, teaching you good habits and workflow along with the regular instruction.

While building some materials for our game, I came across a problem that took me hours to find a solution for- The material blendmode BLEND_Modulate is not supported in the new versions of UDK for lit materials. Long story short, the answer it to use an unlit material. Here’s an example of this in action:

The walls and floor are just the same tiled metal texture. The grime on the walls is a 2-triangle plane with the above material on it sitting just on top of the walls. The mess on the floor is another material set up the same way using a different texture. It’s cool to note that this setup will accept global illumination and shadows properly.

Moving on, I’ve been scratching my head figuring out the best way to build a large warehouse for this level. And I mean Large. It takes up nearly a third of the racetrack. If the movie is to be believed, there’s something like three miles of track winding around inside it.

Originally, I’d blocked the main parts of the warehouse out in max, and was detailing high poly versions of the pieces to use like lego’s and build everything. Since the player is never really going to be looking at the building, and rather just racing around and through it, the High poly baked to Low poly approach was taking way too much time, and still resulting in a ton of assets.  Instead, I’ve gone ahead and blocked out the building using BSP’s. Using some tricks I picked up studying materials and shaders all weekend,  I know that I can make it look as good, if not better, with this approach.

…Modeling. The way I know that I love the career I’ve committed myself to, is that I’m fine blowing off a date to practice and improve my skills. Seriously. Plus, when you get right down to it, GDC is right around the corner. The only way were going to scale this mountain to completion is by sacrificing wants in favor of needs. I can’t wait till it’s done though. Rolling into GDC, or Beer Wednesday, or any industry related event and being able to hand out a completed produce is really is the next level in my mind.

I worked on the Trellis stands that line the track tonight. The sheer scale and height of the object make working on it a bit awkward. I experimented with the sub-object orbit mode to streamline the process.
Here’s a few progress shots. (The trellis Frame is untextured. You’re just looking at the Normal Map and AO.)

Trellis Frame

Detail Shot of Base

Reference Image

Or rather, Box.

Anyways, Boom! Progress is well underway with PDR. A few recent team meetings have really helped, as each team member shares their areas of expertise with the rest.

So, uhm, what am I looking at?
What indeed.

Apparently Terminal Island, home of Death Race, was once a lush forest of rebar and tarnished brass pipes. When they cleared the land to make room for their glorious prison/race track they went ahead and crushed the local vegetation into cubes. Afterwards, they used said cubes to line their track, and to provide shelter to unwitting Pit-crews in times of danger.

Weekends are never a good time for me to get work done, with family obligations, Church, and females begging for my attention. I’m beginning to think the three other members of this project who are all in the midst of long term relationships might be on to something, from a time-saving point of view…

Anyways, I’m about to take this model and a few others which are in the early stages of detailing into Unreal. I think the Spec Map on this particular piece is fighting with the Normal map and becoming nosier than I’d like. I’ll see what softening it up a bit does…

…and I’ve been baking some maps from high-poly models.

I’ve got two weeks off work, and I decided to first celebrate with friends, and then come home and spend the wee hours of the morning doing some test renders of my latest model. The purpose of this particular model was to help me establish a good work-flow. Specifically, how to properly incorporate Low-poly modeling in 3DS Max, High-poly modeling in Mudbox, map making in 3ds Max and Photoshop, and finally importing into the Unreal Development Kit. I’m looking for what areas I end up spending more time than anticipated on, so that I can plan for it in the future, and perhaps alter my flow to be more effective.

This prop is eventually going to be scattered all over the level, but it may be hard to tell just what it is… It’s a concrete footing for the huge trellis’ lining the entire level. It’s a bit like an iceberg, and will be buried anywhere from 25% to 90% in the ground. The idea is by modeling a lot of extra ‘give’ into this piece, it will be much easier to fit the tiling trellis pieces together.

This is a straight Hi-Poly to Low-poly transfer, and a such there are glaring errors in the texture maps. Rather than spend lots of time and computer power re-rendering these, I’m going to quickly clean them up in Photoshop later. Right now the Tri-count is 1900. I’ve got a 500 tri LOD already modeled, and I’ll do a quick comparison in the near future.

The nature of our project lends itself well to lots of tiling assets. The trick to making tiling assets believable, it to make them look like unique versions of the same thing. In the same way that no two manholes on a New York City street would collect exactly the same wear, We don’t want any two tiled assets in our game to look like exact copies.

To this end I’ve been building some simple tests in Unreal to get the most mileage out of our props.

To give you some background on the layout of this test scene, it’s just a bunch of simple BSP cubes. The floor and back wall have a neutral grey material with no specularity applied to them. You’ll see the color of the floor and back wall change though. This is because the level is lit with a single light with Lightmass (Global Illumination) set to 6 bounces. Since the only real color in the scene is red, everything else gets chroma-shifted in that direction.

In the course of putting together this project, I made another personal discovery, Global illumination looks a little silly when the entire scene is dominated by one color (In this case, Red.) In actual application this shouldn’t be an issue.

Finally, here’s the material network I created for this demo. The only thing not shown in this picture are the scale values in the TexCoord nodes. I played with the values to make the grunge appear more subtle, and to make the detail layer fit the block height.