One of the issues that have been discussed for about a year now is the ever important frame rate. 3dfx interactive and their religious zealots would have you believe that high frame rate is the most important trait of any high quality gamer-oriented graphics card. The reality is that it is important but not for the reasons that you think. The religious zealots are only telling you part of the story. In the religious war between high frame rates (HFS) vs. image quality, the image proponents argue that your TV or the movies are only "rendered" at 24 frames per second or 30 frames per second depending on the medium (TV or movies). The argument continues on to point to the simple fact that the action in these scenes are extremely smooth even though the frame rate is so "low" in comparison to Voodoo II. Logic dictates, then, that since the action is so smooth at 30 FPS then frame rates at 60 FPS or higher is a waste. More effort then should be focused on image quality.
The image proponents are correct in their argument. They are correct in their logic. But again, the story is incomplete and, therefore, their conclusion is flawed.
The HFS proponents, on the other hand, think that the argument is baloney and, naturally, do not offer any counter-argument. It is a religion to them. They can only counter that low frame rates offer choppy play and other over-hyped arguments. They even argue that the reason movies and TV seem to be so smooth is because their "effective" resolution is low and that motion blur helps smooth out the action. They even argue that the higher the resolution, the higher the frame rates you'll need to reduce the likelyhood of eye-strain.
The HFS proponents are probably correct in their argument. They are correct in their logic. But like everything else, the story is incomplete. Thus, their conclusion is flawed.
The HFS proponents are routinely amazed at watching DEMOS being played at 100 frames per second. In comparison, watching the same demo at a much slower pace of 30 frames per second seems really dull especially if you have already seen that demo a million times before. But memory is a feature that HFS proponents clearly lack. Actually, to be accurate, their memory seems to be selective. They remember watching the demo at a slower pace so that proves that they do have the ability to recall. While watching a demo at 100 FPS, however, they just seem to forget that they are watching a demo. They are not actually playing the game.
In the real-world, we have to deal with the "mushroom" theorists. Rush Limbaugh explains it best. To paraphrase the big guy, "everything I say is the truth in a political point of view". This means that people are, indeed, telling you the facts but they are not telling you all of the facts. With only a portion of the information to go on, even Albert Einstein would arrive to the wrong conclusion.
For the sake of argument let us say that the scenes you watch on your TV are "rendered" at 24 frames per second. Let us also say that the "effective" resolution of your TV is only around 640x480 (not accurate, but let us pretend). Let us also say that the scenes are helped with motion blur. While watching TV what do you notice?
You should have noticed that regardless of what is going on in the scene, the frame rate is essentially constant. You can have a scene where all you see is a photograph for 30 seconds and another scene with explosions, car chases, and raging flames. They are all "rendered" at the same frame rate. This is what makes the action so smooth. Motion blur helps alot but CONSTANT FRAME RATES are what makes the whole thing look so great.
3dfx interactive recently released a propaganda demo showing off the difference between 30 frames per second and 60 frames per second. In the included text, they inform you that certain parts the scenes are more blurry at 30 frames per second than at 60 frames per second. So, the argument goes, your aiming may be affected by the blurryness.
That is a bunch of horseshit. First, how many gamers out there play Quake II or Unreal using levels that are so huge and your targets are so far away that your target is essentially a small blob on the screen? Second, how many gamers noticed the "blurry-ness" in a middle of a dogfight? Third, how many gamers even noticed the "blurry-ness" until someone mentioned it?
I have played Quake II at 640x480 on a non-mmx P200 megahertz machine. I did not have a 3D card so everything was "in software". My frame rates at that resolution on demo1.dm2 was probably around 11 to 12 frames per second. I am not sure anymore the exact value. By their logic, I must have seen a whole lot of blurry objects whizzing by me. It does not matter if I was playing Quake II on the internet or in single-player mode. Everything that moves must have been blurry. Unfortunately, I know better. And I also have a new machine so I know how the game is like on a more powerful machine, too.
Make no mistake about it, there are blurry objects in the game--as long as that object is in motion. It is present at 30 FPS AND it is present at 60 FPS, too. It is just a tad bit more blurry at 30 FPS. The bottom line is that the "blurry-ness" is so insignificant that it shouldn't be a real issue until all the other problems are worked out. For example, shouldn't one try to make a card that can hold 2048x2048 high-detailed textures?
The demo does show off the point that I have been making in this article--constant frames rates are the key to smooth game play. So the demo is good for something. I have a PII 400Mhz machine equipped with TNT and Voodoo II SLI. At 1024x768 (and my monitor refresh rates set high enough so it is not the limiting factor), I noticed that the frame rates on the "fast" side is not constant at 60 FPS. In fact, the frame rates range from 56 to 66 FPS. This means that the average frame rate is 61 FPS with a variance of just +/- 5 FPS.
While watching the demo at this high resolution, I noticed that the spinning room appears to be slowing down, speeding up, slowing down, speeding up and so on and so forth. The apparent speed changes is due solely to varying frame rates on the Voodoo II SLI. This would affect your aiming far more than any blurry-ness you might see on the screen. Ask yourself this: is it easier to hit a target that is moving in a straight line or is it easier to hit a target that is zig-zagging around? Then ask yourself: is it easier to hit a target that is moving at a constant speed or is it easier to hit something that is changing its speed from time to time?
Constant leads to better performance. It is amusing that I haven't noticed any graphics chip manufacturer or board manufacturer even tackle this issue head-on. Maybe the reality is that this problem is extremely difficult to overcome. Don't talk about it much so maybe people won't notice, is that it?
I also noticed that on the slower 30 FPS side, the spinning room appeared to be slowing down, speeding up also. The "annoyance" factor looks the same on both sides. The difference is that the frame rate only varied from around 28 to 33 FPS. The variance was just +/- 2.5 FPS.
So what does this all mean? Well, it means that at higher frame rates, the amount of variance in the frame rate a person can tolerate increases. Higher frame rates are not better because it is faster. Higher frame rates are better because the TOLERANCE is larger.
We see benchmarks involving average frame rates. That only tells part of the story. What we, the gamers, really need to know is how much the frame rate varies and how often does it vary! A simple number can not give you this information. A graph of how the FPS changed over time would give you this information.
|
[an error occurred while processing this directive]
|
