HDRI

From CGAFaq

HDRI stands for High Dynamic Range Image.

The dynamic range of an image is the range of brightness from the darkest part of the image to the lightest. Since a computer monitor (among other output formats) has a low range of outputs, most computer graphics image formats store a low range of values for each pixel -- for example, a JPEG image stores 256 different values for each color component in a pixel. The real world as we experience it, however, has a much wider range of brightness, from starlight to sunlight. To convert a real world image into a low dynamic range image, the world's dynamic range must be scaled, clamped, or otherwise shoehorned into the smaller dynamic range. This is why an image of the sun and piece of white paper might use the same color to represent both, even though the sun is clearly much brighter.

A high dynamic range image, by contrast (ahem), represents images with a range wide enough that the upper bound is unlikely to be exceeded. Typically each HDRI pixel component is represented as a floating-point value corresponding to some physical measure of the energy in that pixel. HDRIs can be synthesized from real-world scenes by combining multiple photographs of the same subject taken at different exposures. Since HDRIs are a better model of reality, renderers that use HDRIs as the source images for environment or texture maps can generate more realistic images. HDRIs also allow for better post-processing effects, like the simulation of glare.

Once a HRDI is generated, either from photographs or from a renderer, it must be converted to a low dynamic range before it can be displayed. Some conversion methods include simulating a camera's response, simulating the human visual system, or optimizing the conversion to preserve detail.