Summary: A lot of people underestimate the significance of a good technology demo. In Alan's final contribution to the industry, he dissects what the a good technology demo needs to have and shows where the industry has done things right and wrong.
(Alan, we know you're from Stanfurd and all, but since we'd like free physical exams in the future, we wish you the best of luck on your journey to become a physician. -ed.)
In this article, I'll dissect what it takes to make a great technology demo, in the hopes that the future will bring even better product launch demos. After all, competition brought us the hardware we have today hopefully by competition between demo teams will do the same.
SIDEBAR: At Stanford, the introductory CS class holds a quarterly graphics programming contest. There are prizes for two categories: Aesthetic Merit and Algorithmic Sophistication. The year I entered the contest, they ended up giving me an award for Best Combined Aesthetics and Algorithmic Sophistication. It was a cool particle effect thing (given the limited programming environment) that did a starfield screen saver, fountain, etc.
The importance of the technology demo should not be underestimated. Despite all the 'head-to-head' comparisons between ATI and NVIDIA flagship products, the differences aren't as significant as either company would like you to believe. We're not dealing with the era of 3D/fx Voodoo Graphics versus S3 Virge. It's more like Coke versus Pepsi. Sure, NVIDIA has significantly more complex Pixel Shader capabilities, higher color precision and subpixel accuracy but there aren't many real applications that take advantage of it. Likewise, ATI does better with the graphics benchmarks, but at best, it's only a difference of being able to run anisotropic filtering on the X800 XT Platinum and not on the 6800 Ultra. In the end, many people go with the brand-name they prefer and that's where the tech demo comes in. The tech demo is how Sony PlayStation 2 was able to stifle the Sega Dreamcast despite platform parity early on.
SIDEBAR: 'I am.' is the shortest complete sentence in the English language.
On the PC side, because the relationship between software and hardware developers is informal, there is less incentive for software developers to push the limits or draw upon the unique talents of an individual product. The more platform-specific a title, the smaller potential user base there is. With the console, this was OK because it gave developers a chance to reduce their competition (an RPG on the Xbox doesn't have to go against SquareENIX) and because they had an opportunity to make up lost sales through exclusive-platform deals. PC developers face greater competition from other software titles in the same genre that will run on multiple graphics cards, and the amount of support that can be given by the hardware manufacturers is significantly less. Then, taking into account that the average PC gamer does not have flagship hardware (due in part because PC hardware cannot be sold below cost) there is even less incentive for developers to push the limits of PC hardware. Sure a game like Far Cry stresses out most PC's, but we're not seeing games that'll recommend a 3 GHz class CPU and GeForce FX 5900/Radeon 9700 as the minimum spec.
In the end, there are just a handful of developers who have the financial stability and personal passion to create the next world-class graphics engine. The only way to align interests would be for hardware manufacturers to take on first-party software development, a foolish undertaking unless the first-party studio is capable of producing 'quadruple-A' FiringSquad Editor's Choice games at reasonable budgets.
SIDEBAR: Back in the day, there was a saying: 'When Origin releases a new game, it's time to upgrade my PC.' Imagine if one week after the Athlon64 3800+ comes out there's a game with the recommended CPU being the A64 3800 and that the recommendation was accurate and worthwhile.
When developing a technology demo it's a question of whether the hardware manufacturer wishes to target the developer or the end-user the difference between now and the future.
Demos intended for developers need to be realistic. They need to provide real-world examples of new technology features that can be incorporated into games immediately. NVIDIA classically has done a better job than ATI in this realm. With the GeForce4 launch, the Squid demo was intended to demonstrate the use of fluid simulation on the GPU itself. In fact, they were incorporating a fluid model only recently described at SIGGRAPH. Today games like Morrowind and Tiger Woods Golf incorporate on-GPU calculations for their water effects. Simulation of physical dynamics is something NVIDIA seems excited with. From a technical standpoint, 'Nalu' is one of the most exciting demos from the GeForce 6800 demo suite due to the complexity of the hair model and how it moves in water. 'Clear Sailing' features adaptively tessellated waves, dynamic calculations for the foam and splash. These demos provide developers with ideas on how to incorporate GPU-based physics simulation while simultaneously entertaining the end-user. Most NVIDIA technology demos are designed from this perspective. Of exception are the two best NVIDIA tech demos of all time, Luxo Jr. and the Final Fantasy: The Spirits Within. Neither of these are publicly distributable due to the licensing agreements with Pixar and SquareEnix, respectively.
ATI has also incorporated novel SIGGRAPH algorithms in their demos such as their 'Rendering with Natural Light' demo in the Radeon 9700, however in general, their demos have classically targeted end-users more often than developers. Even as far back as the Rage 128, when FiringSquad was just starting, ATI had commissioned 'Rage Dawning.' Animusic was the DX9 demo most Radeon 9700 users ended up showing off. While it certainly highlighted the polygon performance, shadowing, and phong shading techniques available on the Radeon 9700, it was the content that was cool. In order to sync the motion with the music, it's likely that the physics are all pre-calculated and so it's just the job of the GPU to make the graphics happen. Today, with 'The Doublecross,' the animation is motion captured. This not only makes it look more realistic but frees the GPU from physics simulation.
No one expects gameplay graphics to reach the level of quality offered in 'The Doublecross' today's hardware just cannot handle it. However, by showing off what the GPU can do when elements of the animation have been precalculated, they are showing gamers what the games of the future will look like as developers are able to take advantage of faster CPUs.
While developing demos for developers is intellectually challenging and important for pushing software development in the right direction, the best tech demo is still going up against the unaligned interests between software and hardware developers. Even if the technology demo looks great and represents a feasible feature for current games, it will not have a substantial impact on the way games are developed market forces still drive developers to maintain platform neutrality. The idea that a good demo will encourage developers to make more computationally demanding games leading to increased sales of a particular graphics card is flawed. Instead it just preserves an overall consumer interest in upgrading graphics hardware (of all manufacturers) as games progress in their complexity. For that reason, while developer-oriented demos will continue to be important, the emphasis should be placed on user-oriented tech demos.
If technology demos are going to be target end-users, it needs to be visually cool and exciting. It does not necessarily need to represent a plausible in-game scenario it should show a glimpse of the future. After all, when gamers are showing off their new hardware to friends (future customers), they'll turn to the best graphics as opposed to the benchmark. I've given a few technology ideas about pre-calculating physics, but to understand what the art of a good technology demo is, it's easier to turn to real examples.
After watching the trailer, chances are that for the most part, you'll agree with the statement that the trailer was exciting. The point is that demos targeted for end-users should be thought of as machinima. It needs to tell a story. MGS3 has the excitement of top-notch Hollywood action film with 'John Woo moments' and dynamic music. There are exciting characters from the anachronistic flying guards to the bad-ass young Ocelot. There is action and explosions, humor (the alligator head), and drama. Even a demo such as Animusic had its storyline. The characters were the bouncing balls and the story was developed through both music and animation. You wanted to see where the balls would bounce next or what instrument would activate. 'The Doublecross' tells a classic espionage story and leaves the gamer eager to see what happens next.
Important to all three demos were film techniques of editing and direction. That's what a good tech demo needs. It should not be a demo that is left running in the background in a continuous loop like NVIDIA's Dawn. There should be an introduction, middle, and end. People should say 'That was a cool, I want to see that again' as opposed to 'I've seen enough, show me the next demo.' Don't dwell on a particular camera shot to show off the 16-pass shader that took months to code if the scene flows better cinematically with a quick cut -- make the end user want to watch the demo again to get that second glimpse at that effect.
Storyboard the demo there needs to be a screenplay and the recognition of the importance of a director and cinematographer. Bring in exciting characters (even if they're just the inanimate balls) and use unique settings. Spend the time finding a good composer. Even if someone like Klaus Badelt, Hans Zimmer, Trevor Rabin, or Harry Gregson-Williams isn't available, a good soundtrack makes visuals look even better. There's a lot of good source material out there, just watch domestic and foreign movies and anime, or even look at books.
Short animated films as graphics technology demos are not only feasible, but they will be an obligatory cost of new hardware development.
This is my last article for FiringSquad (or any other online magazine). I've enjoyed my time sharing my thoughts on hardware, software, and the industry in general and hope you've enjoyed the articles I've written since 1999. Of course, I'll probably continue to lurk in the shadows and as my profile says:
If you have a problem, if no one else can help, and if you can find him, maybe you can hire Alan...
|© Copyright 2003 FS Media, Inc.|