Graphics Tests in 3DMark05
There are four graphics tests in the new 3DMark that fall in two groups, one of which is limited to SM2.0 and the other needs an SM3.0-compatible graphics card. We’ll describe the tests starting from the SM2.0 ones:
- The first SM2.0 graphics test is a revised 3DMark05 gaming test that used to be called Return to Proxycon. The test scene looks like a typical sci-fi 3D shooter with a troop of space rangers supported with armored vehicles attacking and capturing the space station Proxycon to get some artifact (you can see the full scene with the artifact in the Demo mode). There are now much more light sources than in the original demo (26 against 8); the shadow maps resolution is higher; the scene level of detail is higher, too. The test is somewhat uncommon if you compare it with today’s 3D shooters which but seldom offer you large-scale open spaces and battles. The best illustration of the latter point is Doom 3 where numerous narrow corridors very rarely alternate with more or less spacious rooms. Most sci-fi shooters are like that, but there are a few exceptions. For example, there are large-scale scenes with 200-300 models of enemies in a frame in Starship Troopers.
- The second SM2.0 graphical test is not new, either, being based on the second game test from 3DMark05 called Firefly Forest. Like before, dynamically generated vegetation is the basis of the test. Although the scene is greatly limited in space, the vegetation helps to test the performance of a GPU at rendering shadows and lighting, to test the efficiency of the vertex processors as well as of the central processors of the system J. There is one more firefly in the forest and the shadow rendering method is different than in 3DMark05’s analogous test. The resolution of depth maps and hardware shadow maps is higher, too.
The next two tests use the SM3.0 profile exclusively and thus require a graphics card with Shader Model 3.0 support to run.
- The first SM3.0 graphical test is nothing else but a greatly revised and improved version of the third game test from 3DMark03 which was called Canyon Flight. This test scene makes use of HDR, also when processing reflections and refractions. Like before, depth fog is used under the water surface to create an illusion of depth, but in the updated test the water surface is also distorted by means of two scrolling normal maps and four Gerstner wave functions. The result is very realistically-looking water. Humidity is imitated by means of complex heterogeneous fog. The sky rendering algorithm is more sophisticated, too. The scene still uses only one light source – the sun – but it’s difficult to render dynamic shadows here due to the complex relief of the canyon walls.
- The second SM3.0 graphical test has no analogs in the previous versions of 3DMark. It is completely new and shows how to apply HDR and dynamic shadows in large-scale scenes. The test shows an abandoned arctic station and features a day-night transition during which the shadows are becoming longer. This is a demonstration of the flexibility of the CSM method. The snow uses a Blinn-Phong shadowing model, 2 normal maps and 1 color map, and the subsurface scattering effect which makes it look as real as possible. The test is also indicative of the performance of the graphics card at processing systems of particles which are used in the scene to imitate a blizzard.