• Happy holidays, folks! Thanks to each and every one of you for being part of the Tom's Hardware community!

Importance of Polygons in Today's Graphics

Being the (geek) I am, I was watching a lot of Youtube videos on the development of some old games. In each game they talked about how groundbreaking their 1998 graphics were because they made an engine to render more polygons. Then another guy was talking about seeing polygons from a farther distance. Then another developer was talking about how the sky had 4,000 polygons alone.

Pretty much, in all these 90s games the developers seem to just talk about polygons for graphics. No mention of shaders, shadows, textures, etc. like is heard today. So how important are "polygons" anymore in today's graphics cards, and why are they not talked about anymore since they seemed to be the bullseye of importance back in the 90s?
 
Solution


The increase...
I think techologies like tessellation make polygon counts not only obosolete, but even impossible to make, since the number of polygons varies. Basically, modern games can dynamically generate more polygons for objects that are closer to you, generating more details. Thus, the limiting factor is now on how good the textures are, the quality of lighting, and so forth.

BONUS: Top-secret tessellated toad tech!

crysis_3_cryengine_3_tech_trailer_boasts_top-secret_tessellated_toad_tech.jpg
 


The increase in image quality was substantial as scenes increased from 100 verticies -> 1,000 verticies -> 5,000 -> 10,000 verticies but it has become increasingly less noticeable as scenes have progressed to 50,000 verticies and even 250,000 verticies at the high end.

Right now, increasing polygon count is being looked over in favour of other ways of increasing detail and adding realism.

Normal mapping is an excellent way of simulating an increase in polygon count without actually altering the mesh at all.

Tesselation (a form of dynamic LoD) and geometry shading can be used to fine tune the level of detail dynamically. Meshes that are close to the camera can be algorithmically subdivided to increase polygon count while meshes that are far from the camera can be merged to decrease polygon count. The total vertex count in the scene won't change much, but the verticies in the scene will contribute more meaningfully to the image quality.

Post processing filters have seen a huge increase in use in recent years. Games such as Metro 2033 have used post processing to add incredible effects and immersion.

Lighting is another topic that has been receiving a lot of attention. If I recall correctly, Doom 3 was the first game to use a per-pixel lighting scheme rather than per-face lighting scheme. Since then, HDR lighting, dynamic lighting (as opposed to fixed lighting), volumetric lighting, diffuse lighting, and ambient occlusion have all seen a major increase in usage in high-end titles.

Similar improvements have been made to shadows and occlusions. Take a look at the dust and debris flying everywhere in Battlefield 4 and tell me that it doesn't look epic.
 
Solution