I have been asked a few times why I don't just use a model from a 3d package like Poser for the in game models and such.

The answer is simple: The models made for Poser and DAZ studio ect. are designed for pre rendered movies and artwork, not video games.

The topic today is polycounts or Polygon counts. This refers to the faces, quads or triangles preferably, used to make a character mesh (3d model).

3D model design for CG movies, is a world apart from 3D model design for video games. Movies focus on completely realistic character design and animation whereas video game modeling revolves around building assets that look believable but also optimized.

In a CG movie, once you render out the movie, it is set, done, what you see is what you have and if it looks right, it's right.

In a video game, the models have to be built in a way that the game will run quickly on a massive range of consoles/computers and will stand up to the multitude of ways the models will be interacted with. In other words, Movies = fixed, Games = real time.

There is no polygon budget for movies typically, use as many shapes as you need to make the characters look as great as possible, but in games, you are limited to how many polygons you can use because even though computers are capable of handling a lot more than they used to, having lush environments and dozens of characters running around your game will cause even the best computers to start to lag when there's enough going on.



Take a look at the image where it says "Triangles" that's usually what the game engine will render. If I tried to put a Poser character model into a game engine, I would have effectively used up an enormous portion of my polygon budget just getting the character rendered up in game. If I rigged and then tried to animate the character, all the computer resources would be devoted to rendering that model moving, so it would be the entire game, and that's no good!

If you ever wondered why the cars in Grand Theft Auto 5 don't look as nice as the ones in Gran Turismo 5, it's not because the artists don't know how to model cars well, it's that in GTAIV they have to spread out their polygon budget across, buildings, human characters, items, and interiors, where in GTIV they are free to devote their budget to the cars and tracks.

A typical movie character is modeled in low polygon form first to get the basic structure of the character, the artist then adds more and more divisions until they get the model as lifelike as possible including wrinkles. For a game character, the artist first makes the character in low polygon form, then only adds divisions where necessary to make the character animate realistically, the wrinkles and folds in the clothing are not modeled geometry, but are usually done with texture and 'normal/bump' maps. The normal/bump map changes the way light affects certain parts of a model.


Here is an example how a normal map gets used to give the illusion of detail on a flat surface.

Normal maps are used in very many games now because it's cheaper than modeling out the actual detail and you can get some amazing results.

To sum it up, Poser, Daz Studio, are meant for creating images and movies.

Creating my own models for the game is not only good practice but will allow me to control my polygon budget.