Advertisement

Restrictions 3D Modeling for Gaming

Started by July 22, 2009 04:16 AM
3 comments, last by loufoque 15 years, 6 months ago
Hi all As a part of my thesis I'm making a 3D character that I'll be using in a mini-game (really basic, just to show how the character walks around in a small, basic environment). I've got experience with programming and 3D modeling. But I'm totally new to game development. My programming skills are mainly web & software development (Visual Basic, Java, C#, ActionScript). I've got experience with 3Ds Max and Lightwave but it has always been towards animations/graphics, never for gaming. So I was wondering, are there any restrictions or requirements if you're modeling some character (of any object for that matter) to use in a game engine? I'm thinking about stuff like: - number of polygons (has to be low poly, but are there any guidelines?) - what kind of polygons (tris, quads, n-gons, ...) In my previous experience I always used quads, so I always tried to convert the n-gons to quads, but I have a book on 3Ds Max where the model a game character with a lot of n-gons, so I'm a little confused now. - differences between game for PC, Xbox, PS, ... ? Or is this up to the game engine? - ... It would be great if you guys could give some directions in these things, or some references to articles. It might be silly questions, but I wouldn't want to be modeling for days and have to start all over again, just because I didn't exactly knew what I was doing.
Number of polygons: depends on the target hardware. A single charakter is not much of a burden for any hardware: PC I'd say 10k to 20k triangles, same goes for XBox360/PS3, alot less for Wii, even fewer for PS2, PSP, and whatever more is out there on the market.

That's not an issue for you, though: as a single developer without a large company you won't get developer access to any console except the XBox360. You can develop software freely on PC, and you can develop software for the XBox360 using C# and XNA. But I don't know of any other, although there might be "inoffical" solutions.

The graphics board can handly only triangles. So at one point of the pipeline or another, all quads, polygons or whatevers will have to be triangulated. That's why most modelling programs and exporter plugins offer an option to triangulate the model. I'd suggest using it, and only deal with triangles inside your own program.

Differences between PC and console development - those would fill books. But for your mini-game, none of them matter. Present consoles use GPUs which don't differ much from their PC equivalents.

----------
Gonna try that "Indie" stuff I keep hearing about. Let's start with Splatter.
Advertisement
Hi,

I don´t have any experience with console programming, just PC.

As Schrompf has already mentioned a character with ~20k of polygons might not be a problem at all. In my own "engine" the average number of triangles is ~400000 running at an acceptable framerate.
I think it´s way more important how to draw your objects. In case of having complex shaders even 5000 triangles might cause a crucial framerate drop.
Multipass rendering (e.g. Shadow-Mapping) has a strong performance impact as well.
Switching shaders, states, textures, materials... has to be considered as well, best way to go for that is to group all objects sharing the same "rendering properties".
Another point is you connot use all resources for rendering only as you want to make a game you might want to implement game logic, sound, AI, physics, controls as well.
To get to the point I think it depends on a lot of other things how nice the scene is rendered.


Quote:

It might be silly questions, but I wouldn't want to be modeling for days and have to start all over again, just because I didn't exactly knew what I was doing.


I think that´s not a real problem. I´d rather try to create high-polygonal models in the first step. Reducing polygon count afterwards is not that much of a problem, there are some plugins available which can handle that pretty good. (e.g. polygon cruncher for 3DStudio Max)

Hope that helped a little bit,

Ralph
Polygon crunchers are horrible. Much like a spell-checker is no replacement for a proofreader. They do what they say (ie, reducing polygons) but not in a very good way. They will destroy the topology of your model, and the important edge loops. It's not very useful to work with and rig a model that some polycruncher murdered.

Edward, it doesn't matter how many sides your polygons have when you are modeling. Anything is easily convertible to a triangle. No one works with triangles only while making models. A lot of times, it's a model with carefully placed edge loops that is then taken into Zbrush, or similar sculpting programs.

If you want to be a good game modeler, hang out at the polycount.com forums. That's all they do there.

Since you are familiar with C# already, jump directly into XNA.
Triangles are a thing of the past.
Using polygons with your own tesselator (DirectX 11?) is the way to go as it allows to transfer more compact data, since memory transfer is the real bottleneck with a GPU.

This topic is closed to new replies.

Advertisement