multi-pass or deferred) since we need to invent a clever solution for each effect we're trying to simulate. This is often far more work than just having a unified system (eg. We use smoke and mirrors to achieve both these effects. Reflecting the environment would require rendering a representation of the scene six times into a cube map, possibly for each ship. We also make practical compromises - full scene shadow-mapping would require drawing the scene twice. This is how all of the zone mode effects are implemented. The same system also allows us to intercept and modify parts of the shader that are normally controlled by the art team. WipEout HD uses a fairly sophisticated shader system - we have many possible versions of each shader (not many people may have noticed, but there are real-time shadow-casting spot lights on some of the tracks!) and we really only want to hit every pixel once if possible - this means we need very fine control over how each shader is built since that single pass over each pixel needs to do the minimum work necessary.
#Wipeout hd fury chetcc 1080p#
We render with 2xMSAA in 720p, so there's not a huge difference in memory footprint - actually 1080p needs less memory because we don't have to do the multisample resolve so the alpha-to-coverage buffer isn't really needed.Īlongside the main framebuffer we also use other buffers for post-processing effects which although not at 1080 are still quite sizeable and due to the sheer number of pixels we need to process the fragment shaders need to be highly optimised. Therefore more memory is required for the render buffers meaning we have less available for textures and geometry etc.
#Wipeout hd fury chetcc 720p#
Some of the environments are around 1.5 million polys.ġ080p (1920x1080) vs 720p (1280x720) translates to 2.25x the buffer space required for 720p. Only the ships in WipEout have LOD, and the lowest LOD for a ship is around 10,000 polys (the highest are around 30,000 polys). Because we had existing technology from Formula One: Championship Edition that we'd already spent a lot of time improving, as well as first-class support from our internal technology groups, it turned out we could go quite a long way before we stopped. Because we were essentially iterating PSP assets, increasing texture and geometry resolution over time, the process was really just to see how far we could go, then stop. The challenge was not as significant as it might have been.
0 kommentar(er)
