How are real-time renderers able to render with such quality?

Blouko

Hi,

I was wondering how real-time renderers were able to render scenes so quickly compared to pre-rendered content. As of recently, the latest advances in real-time rendering technologies have made games and movies approach the pre-rendered quality more and more and I still don't understand how they've managed to do it.

So, how is it done, how do they make it possible?

Thanks in advance

PolyHertz

A lot of tricks and cheats basically.

Baked reflections.
Baked lighting.
Low resolution realtime shadows.
Far fewer polygons with detail substituted by normal maps.
Sparing use of unique non-reusable / non-tileable textures.
Limited and low accuracy physics simulations.
Much simpler rigs for animation
etc.

The most noticeable difference right now between the highest end games and film is physics and animation. Water in particular tends to be extremely simple in games, which becomes very apparent as soon as you try to interact with it.

Blouko

PolyHertz said:

A lot of tricks and cheats basically.

Baked reflections.
Baked lighting.
Low resolution realtime shadows.
Far fewer polygons with detail substituted by normal maps.
Sparing use of unique non-reusable / non-tileable textures.
Limited and low accuracy physics simulations.
Much simpler rigs for animation
etc.

The most noticeable difference right now between the highest end games and film is physics and animation. Water in particular tends to be extremely simple in games, which becomes very apparent as soon as you try to interact with it.

I understand now, thank you very much for your reply!

JordanN

GPU accelerated rendering has helped take over a lot of CPU tasks that would take hours to do on its own.

Even on PS3 level hardware, there was a sense of toy story quality because programmable shaders and post processing effects had start entering the norm due to the massive jump in GPU power that the PS2 era consoles and prior where mostly CPU bound.

Obscura

High end real time graphics with filmic post process and advanced lighting tecuniques are still expensive to render, especially on high screen resolution (this is the most critical part as a lot of things are done in screen space, ie post process effect like ray traced screen space reflections, ambient occlusion, depth of field, marching volumetric fog, maybe even GI), but recent videocards seems to be fairly strong, and they can handle fairly large amount of stress very well. Overall scene complexity also plays a part of this as it was mentioned before.

Axi5

approximate approximate approximate

poopipe

Fudged calculations  guesswork and precomputed data make all the difference.