Spider-Man 3 Character Modeling Q&A

With the move to next generation platforms, Spider-Man 3 is easily the most visually impressive title in the series to date, but what exactly is involved in bringing everyone's favourite web slinger - and some of his most memorable enemies - to videogame life? We decided to find out, so chased down Chad Moore, Treyarch's Lead Character Artist on Spider-Man 3.

IGN: How much asset sharing is there between yourselves and Sony Pictures? What's the relationship like between the film makers and Treyarch?

Chad Moore: The relationship between Sony Pictures and Treyarch has been a very good one in my experience. Along with giving us early access to script, Sony Pictures provided Treyarch with both concept drawings and photographic reference materials in order to help us create cutting-edge next-gen graphics. For all of the key characters, we were provided detailed head-scan data so that we could achieve true likenesses.

In order to more accurately represent characters like Venom and Sandman, certain members of the art team (myself included) visited the Sony Pictures lot and spent a day with their computer graphics team. We got to see some early footage from the movie, as well as a number of pre-vis shots for Venom and Sandman. Along with being extremely cool, this visit was invaluable in helping us to recreate these challenging characters.

IGN: How have Treyarch's character models evolved since the last Spider-Man game? (aside from Ultimate obviously since that had its own style). What does 'next-gen' mean for how characters look and move?

Chad Moore: As far as characters are concerned, the single biggest difference between this generation and the last is the number of polygons that are used. For example, the Spider-Man model in SM2 was about 5,000 polygons. For SM3, we used 55,000 polygons… more than ten times the number. Having that kind of power allows us to give characters a greater level of detail than ever before.

Next-gen also means a far more advanced lighting model, which takes advantage of multiple rendering passes per frame. This allows us to use things like normal maps, specularity maps and reflection to make materials look more realistic. Environments can have dynamic lighting, and all of the different surfaces on the characters will react accordingly. The Lizard's skin looks smooth and slimy, while Rhino's hide is thick and rough. It's very cool technology, and brings these characters to life as never before.

Lastly, we are able to support much more complex skeletons, or rigs, on the next-gen platforms. As an example, both Sandman and Scorpion had more than 125 bones and controllers in their skeletons… three times more than we were able to handle in previous games. A more robust rig means better animations and more realistic movement.

IGN: You've got several big bads that won't be appearing in the film, such as Scorpion and the Lizard. How much freedom do you have in creating their 3D model, and what's the approval process like?

Chad Moore: Scorpion is a great example of a classic Spidey villain that we took in a completely different direction. Both Sony Pictures and Marvel thought it was a great idea from the beginning, and the result is a unique and interesting take on an older, more established character.

On the flip side, we had very little wiggle room on a character like Venom. We were given detailed concepts and photo reference, and had to make changes to the character on the fly as he evolved during the movie production. In the end, they wanted him to look as much like the character in the movie as possible.

IGN: Can you take us through the process of creating a model? Where do you start?

Chad Moore: A model starts with reference materials, whether it be an illustration or a photograph (here at Treyarch, we have a pretty sophisticated photographic rig to take our own high-res character reference - very helpful, especially when creating heads). Once the artist has the correct reference, they build the base mesh (or what we call the 'graymesh'), which is an untextured version of the character that will eventually be put into the game.

After the graymesh is completed, the artist will get to work on the normal map using a 3D sculpting package (we use zbrush here at Treyarch). The normal map is used to simulate detail on a graymesh… things like Spider-Man's muscles, or the wrinkles in Rhino's skin, or the folds in Peter Parker's jacket. To put it simply, a normal map makes something with thousands of polygons look like it could have millions of them.

After the normal map is created, the artist texture maps the model. This means creating things like a color map, a specular map, and a reflection map. These textures tell the engine how a character should light within the game environment.

Once the model and texture maps are created, the character is hooked up to a rig and is ready for animation.

IGN: How much of a disparity is there between the models for main characters and for those of civilians on the street - or other less significant NPCs?

Chad Moore: Most of the differences come in the number of polygons - more for main characters, and the size of the texture maps - bigger for main characters. More polygons and larger texture maps means better facial animation, less polygonal edges (read: smoother model), and better lighting on the character. As these characters will often have close-ups during cinematics, we need them to have a higher level of quality than less significant NPCs, like pedestrians.

IGN: How closely do you work with the animation department?

Chad Moore: We collaborate daily with the animation department, as great characters are the result of the 3D model, the rig to which it is attached, and the animation of that rig. If a character's mesh is not deforming well during certain movements, then the animators may ask us to make an adjustment to the model.

IGN: What is holding back in-game photorealism for character models? Is it the resolution of the textures? The complexity of the polygonal model? The shaders?

Chad Moore: For any game, it is really a combination of all three. Making a completely photo-realistic character within a game environment is becoming possible, but having to share computational cycles with other important things (like all of Manhattan Island in the case of Spider-Man 3), means that compromises are inevitably made. And there are still a few technological hurdles that have yet to be overcome in real-time environments… things like ultra-realistic hair and cloth, or true skin and muscular deformation, which really aren't possible yet in games. But we're working on it!

©2007, IGN Entertainment, Inc. All Rights Reserved