If you can't wait until Prey is finally released, then you're not alone. In the meantime, feel free to satisfy that first-person-shooter jones with our sixth Prey Designers' Journal. Meet the Prey design team Previous Prey Journals • #1: Into the Fire • #2: A Texture Is Worth a • Thousand Words • #3: The Numbers Game • #4: We Are the Hive • Mind • #5: Preditor or Prey? by Allen Dilling, David March, and Paul Schuytema (11/13/98) "The polygons are alive! Alive, I tell you!" Remember Dan Gorlin's classic game Choplifter? That was one of my all-time favorites, and one that I still fire up on my trusty MAME emulator. One thing that made that game so absolutely cool was the chopper...if you changed directions, you could actually see it turn. It looked like that chopper had actual mass and volume because of the way it swayed left and right. Of course, we're more sophisticated now, and it takes more than a bunch of 16-color pixels in an animated sprite to give us the feeling that there really is depth there. As long as we're reminiscing about old-time games, do you remember which early arcade game boasted the first real 3D graphics? There were a few early ones, but the granddaddy of them all was Atari's Battlezone--it was the first "true 3D" game to ever see production. Red Baron, another 3D game from Atari, was in development at the same time--and it was a heated race toward completion. Both games worked off of the Asteroids codebase, and Red Baron was the first with 3D graphics, but the "bottom up" approach that the Battlezone team adopted proved to be the wisest decision, and Battlezone beat Red Baron to the arcade by several months. Eighteen years ago, when Battlezone premiered in arcades, there weren't any 3D game artists out there. Zero. Zilch. The Battlezone team hired 3D industrial artist Roger Hector to craft the game's dynamic 3D objects. Hector had done 3D CAD work before, but had no experience with real-time 3D. Hector didn't have the luxury of using any sophisticated tools, but he did have the peace of mind that came with knowing that the two tools he used, a pencil and a pad of graph paper, would never succumb to a "fatal exception error." To create the models for Battlezone, Hector drew the objects on graph paper, very precisely and from several different views, all with precise coordinates. Then it was Ed Rotberg's task to key the coordinate info into the game's world database. With any luck, and a lot of trial and error, they would eventually see one of Hector's tanks "alive" in the luminescent green of the game screen. Times have certainly changed, and this industry now employs hundreds of 3D modelers. Gone forever are the days of graph paper and mechanical pencils. Now, tools such as Wavefront, SoftImage, and 3D Max dominate the industry. Modelers now sculpt their 3D creations in the same "virtual ether" that their finished works will inhabit within the game. In Prey, such components as weapons, enemies, friends (far too few for our hero Talon's liking!), and vehicles are crafted via the creative necromancy of our two modelers, Allen Dilling and David March. What do David and Allen have to say? David March, 3D modeler/animator: First I try to get crazy ideas about what the model will look like. I think about functionality and what the model will do--what it will look like accomplishing its tasks, functions, and so on. Then I sketch. Once I have a big trash can full of sketches, I've hopefully got something that the team just might go wild over! These may not even be good drawings at all, but they serve the great purpose of nailing down the general feel of the model. After I get some kind of conceptual feel, I refer to my sketch while I'm modeling. It's also much faster to grab a piece of paper and start doodling when I get a good idea. When I'm modeling, I'm always seeing cooler ways to do something or something that I could change on the fly. Then I grab the sketch pad and start doodling another arm, leg, helmet, whatever. Allen Dilling, 3D modeler/animator: The process for planning a model can come from a few different areas. Sometimes you have a great visual in your mind about a really cool- and evil-looking enemy, and that can make designing it easier. Another path, which is the way I usually go about it, is to brainstorm the gameplay elements of the character first, and then build its look and feel around that. I talk to the GSEs [Gamespace engineers] a lot and see what they might want in an enemy--to help with the levels--and then I try to fill a need for the gameplay. It's easy coming up with cool-looking stuff; the trick is to make it play well and actually be fun and challenging to fight against. After getting a rough concept in my head about a character, I'll take a few hours and draw some quick sketches of possible looks for the model. Sometimes I'd really like a certain detail in one enemy, but it wouldn't get used for some reason, so I'd dig up my old drawings to comb them for things I could incorporate into new designs. I don't believe in saving cool ideas for the next game, I'd rather go full-bore right now and not save my punches for later. Model to the MAX Once an idea gets the thumbs-up, then it's time for the actual modeling. We use 3D Studio MAX for Prey modeling, and it works well for our needs. All models still need to be converted into "Prey actors" before they can show up in the game, and for that, all work is exported into VRML (that also means that gamers can use whatever tool they want to create their own 3D content for Prey--they just need to be able to get it into VRML). Allen Dilling: Once I have a rough sketch that's close to what I want, I think the character through in terms of how it should function, which weapons it might use, the environments it will interact with, and so on, and start modeling in MAX with all these factors in mind. I'm a disciple of form following function, so I try to make the character really work inside the Prey universe. As I start to model it, I see different things that would make the character better, and by the time I finish the model, most of it has changed quite a bit from the original sketch. I feel sorry for modelers who have to make an exact replica of another artist's concept drawing; they're really missing out on a lot of creative freedom, not to mention fun.   Blow it up!      David March: I always begin with the head. I am just magically drawn to this feature every time. I then model the body and finally, the appendages. Then come the little things; horns, antennae, claws, and so on. I like working with cylinders the most. I use spheres for heads and boxes for feet and hands. I also model using more polys then I need for certain things. It's easier to cut down than to add. Bone This Once a model has been sculpted, the next step is to create the skeletal system. These virtual bones are what are actually animated--each of the model's vertices are associated with one or more bones. When the bone moves, the model's mesh deforms along with it. Boning a model is not only tedious, but it's a great way to showcase flaws in the model's design (a leg may not bend properly, for example). Because of that, the modelers bone a model and create at least one moving animation first before sending the model off to be textured--that way they can catch the majority of the joint problems before it becomes a hassle to fix. Allen Dilling: There is nothing really bad about modeling, but the semi-crappy parts are mainly just doing the technical stuff, like boning the model, and also making sure everything lines up correctly so it will all fit together once it's imported into the game. When boning a model, you need to consider what movements the character will have and also how the animations will affect the model's shape. Assigning most of the model's polys, such as the arms and head, is pretty straightforward, but I usually experiment a bit with the waist and groin areas. These are affected the most because the models are going to be walking and running most of the time. They need to look good and avoid warping too much during the animations. David March: Bone weighting sucks the most! This can be very tedious at times.   Blow it up!      Don't Go Tryin' to Change Me... In a perfect world, when the modelers finished up a model, they could just move on to new work, and never have to revisit their old models again. Unfortunately, that never happens. Sure, we may think the concept for a model is right on, and then we take a look at the initial model and give it a thumbs-up, but there are so many variables at play that the first pass never nails everything right on. Nearly every model has to be reworked in some way. David March: If there are enough polys in a certain area, I find it generally easy to manipulate and change models--pushing and pulling vertices--on the fly. After something is "boned up"--vertices are assigned to the internal skeletal system--I find it harder to change things. You can still do it, but not nearly as easily as before. Allen Dilling: Almost every model gets changed over the course of development. This is and can be a pain in the neck. Sometimes it's only a little thing, such as adding fingers or reducing ploys on a character's butt, and that's not too hard. The main problem is that with our system, you have to rebone the model in Bones Pro [a character animation add-on for 3D Studio MAX] and then export the necessary files for the engine to read. This ends up happening again, and again...and again. Overall, however, it's a small sacrifice considering the kind of dynamic characters we can control with our engine. I'm really glad to be doing it this way and not having to deal with framed models. Talkin' Textures Once a model has been built and boned and a few animations have been crafted to verify its design, then David and Allen export their work to VRML and hand it over to senior artist Steve Hornback or artist Scott McCabe to be textured in Skinner [an in-house tool for putting textures on characters, previewing animations, and so on]. Allen Dilling: I give Scott my original drawing of the character so he can get some details from it. However, if it's a particularly rough sketch, then I just point out some things on the model that need to be done a certain way (like alpha channels, environment mapping, fractals, and so on), and then I explain what the model does (he can watch the animations for reference) and let him go crazy with it. After working with him for over a year and a half, I feel very comfortable letting him skin the models. His work is always impressive and ultrarealistic. The most bone-headed mistake I've made in the past year and a half was trying to texture my own models when I first got hired. I had never done a model skin before, and they were nasty to say the least. David March: I try to model as texture-friendly as possible. It takes a lot of practice to figure out how to make a model work with the textures...it can still be a gray area sometimes. We also give the artists texture poses to help them out--we try to stick the model in a good position to access all the faces for mapping. Sometimes that's not possible. And yes, we do hear "This doesn't work--you'll have to do it again" from time to time. That's when I go postal! What, Me Mo Cap? Thinking up the shape of a creature and then modeling it requires a boatload of creative skill, but the most demanding task, by far, is creating the animations themselves. Nothing breaks the illusion of reality faster than a slinky beast duckwalking as it stalks you. The animation needs to be fluid and natural, and it has to feel right...Sometimes that's a tall order when you're animating a creature that's never been seen before. David March: How do I capture movement? Study, watch, and learn. Subtle movement is the key. I am always trying to learn and watch. We get up a lot and run around like idiots. Creativity helps with odd-looking creatures. Weight is hard to capture as well, but when it's done, it's the icing on the cake. Allen Dilling: Creating organic motion in the animations takes a lot of thought and research. We have some books about motion and various animals' movements that are helpful, but the key is just using your head. When a character is moving, you have to think about all the factors that are involved in the motion. How does the character's weight affect the bending of the legs? How much force is being applied to the collarbone? How would the hair swing during motion? Everything should make sense. A good test is to do the motion yourself and analyze what is happening to your own body. I even bought myself a big mirror for my office to check out my body while I try different movements. Sometimes I try it by myself with the lights off...wait a minute, did I say that out loud?!! D'oh! Personally, I can't stand motion capture in 3D shooters. I think it sticks out like a sore thumb, and if you let some actor and a computer do all your work, what does that leave for you? Maybe I'm just old-fashioned, but I still think of animation as an art form. It's a skill that must be learned through thought and experience. Plus, if you need mo cap [motion capture] for an FPS, then you probably need better animators. Now, I'm not totally against mo cap for all games. I love a good RPG or cinematic that uses it to get some great results, but I think that with today's engines and gaming needs, a good artist can still give you what you need. Also there's a lot of data that you have to filter through to pull out the frames of animation that will interpolate smoothly together and if you want to change the motion or create a new one, you have to schedule a session and do all the crap again and again. OK, OK. I'll get off my soapbox now. Bang! You're Dead! In addition to crafting the enemies for Prey, David and Allen also craft the tools that let you destroy those enemies--the weapons. Weapons are a lot like characters--they too need personalities. Allen Dilling: Designing weapons is the most challenging part of my job. More than any other model or character we make, they affect the gameplay and overall fun of an FPS. You have to be very careful and not just create a bunch of BFG-type badass weapons that will destroy the universe. Balance is the key to successfully designing weapons and power-ups. This is especially true for the multiplayer aspect of the game. The whole replay and longevity of a title depends on having fun weapons. With Prey, the whole team has input into a weapon's creation. We want stuff that will work with the game as a whole and enhance the rest of the experience. Actually modeling a gun is a fairly quick process if you already know what you want. It's a rush to get an idea, model it, do some coder stuff, and then a day or two later, be killing someone with it. Plus being able to test the weapons is ultra important so you can tell if it's fun and if it looks great in the game, too. David March: It's fun to make weapons, and they're not nearly as hard to animate as the organic creatures. Concept and functionality are the hardest part. It goes about the same way as starting a character, but with weapons, I can just jump right in and start modeling with an idea. It's a Wrap! Even with Prey's dynamic environmental geometry, the game world can still seem mighty sparse and desolate if there are no creatures populating the shadowy nooks and crannies. It's been said that fiction is basically about characters interacting--all the setting and plot in the world is there simply to facilitate the character interaction. That holds equally true with games--we play because we want to interact, and we want to interact as characters with other characters. For a 3D action game such as Prey, those characters are the virtual sculptures of our modelers. Working like some strange cyberpunk Michelangelos, they sculpt their creatures from mathematical points in space, threading vertices into triangles, triangles into shapes, and shapes into creatures. But sculpting is not enough--those shapes must come alive. True, during the course of play, the creatures are controlled by computer intelligence or another networked player, but all that you see--every fidget, muscle flex, and war cry--has been animated by Allen and David. Now...if they could do all of that on graph paper, then I'd really be impressed!