A Few Words on Polys and Portfolios

by Darius Kazemi on April 29, 2010

in art,portfolios

I’ve run into some interesting misconceptions from students during my month-long residency at Full Sail. I’ll try to address these on my blog over the next month, this entry being the first.

Team-based work in portfolios

Here’s a common question I get from student artists: “If I worked on a team project and contributed a small part of something, should I put that in my portfolio?”

It’s an understandable question but there’s a simple answer: yes you should put that in your demo reel. All you need to do is put a note on it saying, “Textures by [your name]” or “Enemy model by [your name]“. If it’s a reel it should be a caption at the bottom of the screen during that portion of the reel; if it’s a screenshot for a portfolio gallery just put it as a caption at the bottom of the screenshot.

Students sometimes think it’s weak to include something that you only did a part of in your portfolio. I laugh when I hear this. When you’re working at a game studio you’re only going to be contributing parts of things to a team effort. By including examples from projects where you’ve done a small part, you are demonstrating to me that you can work as a team. I’d rather see that you can contribute to a team than that you’re capable of spending a lot of time putting together an awesome high-poly model.

High-poly vs low-poly

Which brings me to another topic: high-poly models. Some students love having high-poly models in their portfolio. This is misguided. Because we render in real-time, game development is all about efficiency and faking things. The most impressive skill that you can have as a modeler is the ability to create a low-poly model that looks really good.

The majority of game developers aren’t working on Gear of War. We’re working on Facebook games, Unity games, iPhone titles, DS, PC casual, etc. In those cases you’re going to be forced to use low-poly models. A high-poly model is literally useless for an iPhone game.

The most impressive skill that you can have as a modeler is the ability to create a low-poly model that looks really good.

Even if you are working for a big-budget AAA title, consider this: most games have something called a “poly budget,” which is a rough estimate of the total number of polys that you can render at once. This varies depending on your game engine and your platform. Sometimes you can get away with a lot fewer polys if your game engine does a lot of normal mapping and lighting tricks. But the important thing to know is that every game has to work within a budget. If you can take a 10,000 poly model and make it look 85% as good in 5,000 polys, we can now render two of those ¬†models on the screen for the same poly budget. That would make the difference between having 3 enemies on the screen and 6 enemies on the screen!

The important thing you should take from this is that good-looking low poly models are extremely useful in game development, and probably more useful than high-poly models. I’m far from an expert on exact details for this kind of stuff. Fortunately, game industry artist Rick Stirling has some great articles on his blog about this topic. I recommend you read his answer to the “How many polys in my game” question and also his list of poly counts for characters in major games from the last 10 years.

{ 3 comments }

SJML April 29, 2010 at 10:39 am

One of the things I like to see most in a demo reel is a rendered model and then a wireframe view, so I can get a sense of how much detail they can provide in a tiny amount of space. (Displaying a numeric polycount is also nice, since it can be tough to eyeball.)

NSH April 29, 2010 at 11:04 am

I think keeping poly-count down is less important nowadays as it once was. Obviously demonstrating you are aware of triangle budgets and knowing how to use triangles efficiently is important, but often nowadays the art pipeline consists of making a high-poly model, running it through a tool chain to produce a lower-poly version and corresponding normal/height-maps, and hand-tweaking the resulting geometry.

What I would ideally like to see in a portfolio is both a high-poly source model, and a side-by-side view of a simplified version that shows that you know how to extract the maximum detail from your budget. If you’ve got a 5,000 triangle model sitting next to a 500 one that looks almost as good, you’ve demonstrated not only artistic skill, but proficiency with your tools and a knowledge of the domain and practices as well that some Technical Artist isn’t going to have to spend man-hours teaching you.

Darius Kazemi April 29, 2010 at 11:06 am

Agreed — but I’m more responding to students who think that just showing a 40k poly model is the way to go. Definitely showing a low and high poly version of a model side by side is a great thing to do. Also good low poly models are still extremely important for lower-tech games you’d find on the web, DS, etc.

Comments on this entry are closed.

{ 1 trackback }

Previous post:

Next post: