Recovering High Dynamic Range Radiance Maps from Photographsversión en español
IntroductionFilm cameras were developed in order to record light so that it could be reproduced on photographic paper; digital cameras so that it could be reproduced on a computer screen. Neither computer screens nor paper can display nearly the dynamic range (ratio between dark and bright regions) as what is present in the real world, and as a result cameras are not designed to capture even close to such a range. However, by taking a series of pictures with different exposure settings the range can be covered. With this technique such a series of images can be combined into a single high dynamic range image called a radiance map. Radiance maps are useful for representing true illumination values in image-based rendering applications, and are useful for recording incident illumination and using this light to illuminate CG objects for realistic compositing (see the SIGGRAPH 98 paper on image-based lighting.) Recent work has proposed techniques for using high dynamic range images as texture maps in realtime graphics applications (see the EGWR 2001 paper on Real-time High Dynamic Range Texture Mapping.)
Radiance MapsDownloadable in Greg Ward's RADIANCE red-green-blue-exponent format:
Light Probe Images
Software for creating and viewing HDR imagesHDR Shop, a high dynamic range image processing and manipulation package developed at the USC Institute for Creative Technologies, is available for download at: http://www.debevec.org/HDRShop/. It is free for academic and non-commercial use and is available to be licensed for commercial use.
These images are in the RADIANCE Synthetic Imaging System image format (Described in Greg Ward's "Real Pixels" article in Graphics Gems II.) They can be viewed on an X11 display using the ximage program and/or converted to floating point data using the "pvalue -df -H -h" program from the RADIANCE package.
Related ResourcesWhile working at SGI, Greg Ward developed two new high-dynamic range encodings for the TIFF image format: "LogLuv Encoding for TIFF Images".
Slides from Greg Ward's 5/21/2001 talk on "High
Dynamic Range Imaging" at the USC Institute for Creative
Technologies are available here as Ward-HDRImaging-20010521.pdf