| News | Articles | Ask a Question | Question Archive | Downloads | About |
How a Graphics Card Works - Page 13
|
End Frame
In the early days, graphics cards were simply translators, taking the rendered image created by the CPU, and then converting it to the electrical signals required to display the image on the monitor. This worked, but all of the processing was done by the CPU, as well as the player input/output, sound, and any other processes running in the background (such as virus scanners). As time has progressed, a large proportion of the graphical rendering work has been offloaded to the graphics card itself. The demand for more realistic games with higher frame rates has now pushed the technology inside graphics cards to previously unimaginable levels of complexity, indeed, a modern graphics card is today more complex than a whole computer from not too long ago. I've only touched on some of the more complex concepts here, as you could quite easily dedicate an entire website to just one or two of the subjects covered in this article. Hopefully this article will have given you a basic understanding of some of the underlying mechanisms at work while you're playing your favourite games (and maybe you'll shout at the poor little thing a little less when it's creating 50 frames per second in F.E.A.R now, well, maybe not..). So what does the future hold? For most people, photorealistic rendering at full speed is the goal, and human skin and facial expressions are the most difficult to create, but reality is a lot to live up. It may occur to you that there are films which boast more realistic graphics than ever (Shrek, Monsters Inc.), so why can't you do that on your PC? The answer is simple - rendering times. In a Hollywood blockbuster they can afford to wait over 20 minutes for a single frame, so having an insanely high polygon count isn't really an issue. On a PC the average rendering time for a frame needs to be around 30 milliseconds for smooth animation, with more powerful cards getting to under 10 milliseconds in some cases. Waiting a fraction of a second for a single frame can feel like an eternity when playing a fast passed action game sometimes, just imagine several minutes!
For truly photorealistic gaming we might need polygons the size of pixels, a few hard disks full of textures, and graphics cards that are capable of handling it all fast enough to keep the animation smooth, but sadly that hardware is still a while away (you won't be rendering the likes of Monsters Inc. in real time, or confusing your display for what you can see out of the window just yet). But, programmers are creative, and are always finding 'tricks' to help improve image quality, while reducing the processing overhead needed (such as Normal Mapping, or techniques which take advantage of the many fallacies found in the human visual system), so 'almost' photorealistic gaming might not be very far off. Even current games are getting fairly realistic, take this example from Doom 3:
It's not perfect, but it's certainly a step up from some of its ancestors:
An unfair comparison I know, but it really goes to show how much things have improved in the last few decades. To finish off, I shall leave you with an image of a character from the upcoming Unreal Tournament 2007 game, rendered on the new Unreal 3 engine. Thanks for reading, and come back soon!
If this article has got you interested in 3D graphics, and you would like to learn more, then it certainly wouldn't hurt to check out some of the following links:
Article Navigation: |
|
© 2005-2008 ithinkibrokeit
Disclaimer | Contact Webmaster