Optimizati On 18% Gray Cards and Digital Photography Our recommendation is to use either a Gretag Macbeth color checker or a Whi-Bal card for determining accurate white balance. 18% Gray Cards were designed to assist in establishing proper exposure for analog photography. 18% Gray Cards do not provide sufficient data to deter- mine an accurate white balance and they may not be spectrally neutral. maXimizinG imaGE QUaLitY Demosaicing Factoid Many different demosaicing algorithms are in use. New ones are being developed on a regular basis, although most of the new versions are modified or blended ver- sions of the original algorithms. The fundamental goal of a good demosaicing algorithm is to preserve detail and minimize artifacts. Some algorithms are less complex than others and require less computational power. The trick is to find an algorithm formula that provides high quality but is not overly complex. Most commercially available parametric image editing applications make a calculated trade-off between image quality and compu- tational complexity (speed of processing). It can be inter- esting to study comparisons (http://www.rawtherapee. com/RAW_Compare/) of the various raw demosaicing methods. Some parametric image editors, such as dcraw, often take much longer to process a raw file because they are expecting that their user base will tolerate slow pro- cessing if it means the highest possible quality. 156 A hot topic these days is how well parametric image editors handle raw files that were shot with extremely high ISO settings. Demosaicing algorithms can be com- bined with denoising algorithms to combat noise, although this is more often done as a separate antinoise filter step. Third-party plug-ins are often used for the best results, for example, Nik's Dfine, Noise Ninja . . . Since some camera makers control noise more effectively in JPEG file capture (using noise reduction settings) than they do with their raw processor, there is clearly room for improvement. Because noise, sharpening, and saturation are all interrelated, understanding how best to use these functions for high ISO images is critical to achieving the best image quality. Some studies have been done indicat- ing that using one kind of demosaicing algorithm on low ISO images and another one that combines denoise algo- rithms on high ISO images would be ideal. Obviously, this adds complexity to the software, especially when you consider the number of raw file formats and camera models to support. A sensor can be profiled for noise characteristics just as it can be profiled for color response. Another area ripe for profiling is lens characteristics, spe- cifically aberrations, vignetting, softness, and distortions. DxO software has specialized in this area for some time, essentially profiling a growing list of commonly used lens and sensor combinations and developing algorithms to correct for these faults. TripWire This is a good time to discuss the best way to evaluate image quality, particularly with regard to image noise, aberrations, focus, level of detail, and the like. All of our