The hidden story of the 3D engine | Hacker News Reader

The hidden story of the 3D engine

guardian.co.uk

Tim Sweeney raises some PL flamebait: -- "Are games at the forefront of physics and computer science research or forever lagging behind academia?

Sweeney reckons it's a bit of both: "Game development is at the cutting edge in many disciplines. The physics in modern games includes rigid body dynamics and fluid simulation algorithms that are often more advanced than the approaches described in research papers. Over the past 15 years, games have caught up with 40 years' of graphics research and are now leading the charge in some domains, such as practical lighting and shadowing techniques.

"In other areas, we're still stuck in the Stone Age due to ingrained technologies. The C++ programming language, used in all modern games, was hastily conceived in the 1980s as an extension to the 1970s C programming language. Many of the problems that plague computers today - security vulnerabilities, viruses, and so on, can be traced to problems in this language."

Not even a nod toward id Software. Sad.

Also, I have to know, what is "Scottish indenting"?

If I had to guess I would say one space for each level of indention.

Super article. The only thing it left out was the fact that the unreal 3d engine is now freely available for non-commercial use.

The article seems to assume somehow that photorealism would be the pinnacle of game engines. But I disagree. I feel that games are steering more and more away from photorealism towards more artistic direction. Same way that painters realized that creating photorealistic picture wasn't the point when painting, game developers will realize the power of nonrealistic graphics. That doesn't mean that future engines wouldn't continue to advance technically, but rather that engines will allow artists/designers to express themselves more easily and freely.

But who will be the Picasso of game development, that shall be seen. Somehow I just doubt (s)he will come from the EA creating yet another annual Madden or "Generic Shooter 3: More brown lens flare".

Photorealism is technically difficult to achieve, yet easy to judge. That's why it's commonly used as a graphics benchmark. Also, photorealism and artistic direction are not mutually exclusive. After all, photography is an art form too.

Check out Love if you're interested in impressionism. http://www.quelsolaar.com/love/

I wouldn't quite say that a 3D engine is "beautiful," in the normal sense of the word. Raytracing is beautiful—a 3D engine, meanwhile, is eminently practical, an exercise in complex, layered optimization.

I would say that depends on your definition of beautiful. Are you saying that games based on raster-based 3d engines aren't beautiful? If so, then I quite strongly disagree.

I never said that the output of the engine isn't beautiful. I was referring to this line, early on in the text:

> But here is the minor tragedy at the heart of modern games: no matter how astonishing they look, players will never see one of the most beautiful components: the 3D engine.

To say that the 3D engine is "one of the most beautiful components" of a game, is similar to saying that the optical fiber interconnects are "one of the most beautiful components" of a server rack. Sure, they're necessary, they solve the problem well, and there's some neat math involved in making them work as fast as possible, but there's nothing aesthetically pleasing about them, especially if you think about having to maintain them or make changes.

The naïve ray-tracing algorithm that you can sketch on a napkin and write in 20 lines of Haskell? Practical ray tracers - those used for real-world applications such as rendering movies - have much in common with game engines. Both are large and complex systems of many, many modules and employ an array of cutting-edge algorithms (some modern game engines incorporate ray tracers, and today's renderers employ much more than ray tracing). Both are designed with a focus on performance and robustness, while incorporating the large toolbox of features demanded by content creators. They are machines that have gone through many, many revisions of architecture, and have decades of performance tuning and wisdom embedded in them. And they are beautiful in the sense that they have tamed a real problem and exposed its solution in a simple manner, in spite of the complexity they must encompass.

The former, actually :) Having not looked at the field in a [large] number of years, I hadn't realized that there were "practical" ray tracers; I was trying to use them as an example of something that is pure, naive, elegant, simple, and completely impractical; the opposite of a game engine in all respects. (Pretend this is 1995; it would make more sense then.) A practical ray tracer is, for all intents and purposes, just another 3D engine, in the way I was using the term.

What you have said, I completely agree with. However, your conclusion, though correct on its own, doesn't validate the text of the article. Let me quote it again:

> But here is the minor tragedy at the heart of modern games: no matter how astonishing they look, players will never see one of the most beautiful components: the 3D engine.

This line implies that it is the internals of the 3D engine which are beautiful, not their exposed API (because the user may well run into that, in the course of, for example, writing a mod.) As I understand aesthetics, "beauty" is the quality of a work that allows it to be appreciated by those unversed in the art or craft that created it. No one but those versed in the CS, maths, and engineering of a solid 3D engine would dare to call the internals of one "beautiful" upon inspection, so I believe my previous statement holds: a 3D engine is not beautiful.

Why unversed in the craft? Is that a common definition of beauty? That way, no code can be beautiful, right?

The article raises the tantalizing possibility that an Unreal descendant running on future hardware could be the basis for something like The Matrix.

Reply on news.ycombinator.com