High dynamic range rendering

from Wikipedia, the free encyclopedia

High Dynamic Range Rendering ( HDRR ) describes the synthesis of images ("rendering") in computer graphics , taking into account the large fluctuations in brightness that occur in nature. In contrast to the conventional 256 brightness levels per color channel , with HDRR colors are represented internally with sufficiently high precision to cover a very large range of brightnesses. This enables the representation of strong contrasts without excessive loss of detail and the application of effects such as the simulation of lens scattering .

A special technique from the HDRR area is image-based lighting , in which a computer graphics scene is enveloped and illuminated by an HDR image. This creates the impression that the artificially modeled objects are inserted into a natural environment.

HDRR in real-time rendering

With hardware-supported HDRR real-time rendering , the graphics card processes color information internally in such a way that a large brightness range is covered and that rounding errors are kept to a minimum . For this purpose, colors must be represented with a higher storage space. If the accuracy is too low, the color transitions might become so coarse that this would result in visible artifacts (color gradations).

The result of the HDRR process is an HDR image , the dynamic range of which has to be reduced using tone mapping for display . Since tone mapping tries to retain image details, scenes with high brightness contrast can also be displayed without very light or very dark areas being limited by a small range of brightness and appearing as pure black or pure white. An example of a scene where the benefits of HDRR can be seen is a dark cave leading into a sunlit outdoor area. The tone mapping operator adapts to these different brightness levels so that the absolute brightness values ​​of the scene are optimally transferred to the low dynamic range of the screen.

HDRR as a technology for computer games became available around the same time as DirectX's Shader Model 3.0. However, these two technologies have little to do with each other. B. shows that theoretically every DirectX -9 graphics card is able to display the HDRR effects: Every DirectX-9 graphics card must support at least 24 bits per color channel internally in the pixel shader ("FP24") and textures as well Can read 32 bits per color channel ("FP32"), but this would lead to significant performance drops on older cards. The hardware of the 7th generation of game consoles such as the GPU of the Xbox 360 developed by ATI or the NVIDIA GPU of the PlayStation 3 also support HDRR.

HDRR is used in the Unreal Engine 3 , in the first-person shooter Far Cry from version 1.3 or in some maps in Counter-Strike . HDRR effects are used very clearly in the racing game Project Gotham Racing 3 and in the first-person shooter Halo 3 .

Image-based lighting

With image-based lighting (IBL), a scene is completely enveloped by an HDR environment map, also known as a light probe . An HDR Environment Map can be generated by recording it directly with special cameras with rotating lenses or fish-eye lenses . Alternatively, a sphere reflecting the surroundings can be photographed or several individual photos can be stitched . If necessary, the image is then transformed into a cube or sphere that encloses the scene.

One of the problems that must be solved with IBL is to sample light regions of the environment map preferentially in order to enable importance sampling in Monte Carlo ray tracing . Ideally, the scanning positions are distributed in such a way that they have a large distance from one another, but still do not reveal an ordered pattern ( blue noise ). Very bright, concentrated regions such as the sun not only require special techniques for the recording, but are also removed from the environment map and explicitly modeled, since otherwise there would be very strong noise (individual bright pixels ).

IBL finds an important application in film effects, in order to show artificial objects and beings as if they were actually there when the picture was taken.

literature

  • Erik Reinhard et al: High Dynamic Range Imaging, pp. 367-462. Morgan Kaufman, San Francisco 2006, ISBN 0-12-585263-0

Web links