Digital photography has now come to include a digital/digitized image at any stage of production. For film photography, it has come to include the digital darkroom, comprising the scanner, computer with processing software, and printer. As this is a developing technology, it is less well understood in user land than some users find acceptable, and so discussions about equipment and usage thereof continue unabated.

Of interest in the Usenet newsgroup comp.peripherals.scanners is how much information may usefully be extracted from a scanned film image. As scanners become more sophisticated and capable, the digital information format becomes an issue. For consumer scanners, color is recorded as red, green, and blue (RGB), and each color has its own "channel", where information about each pixel is recorded in a single digital "word". The question is: how big does that word need to be to contain all the useful information.

The traditional word length has been set as multiples of 8 bits. Thus, the common "word length" are 8 bits, 16 bits, 24 bits, 32 bits, and so forth. An 8 bit word can record 256 different values (2^8). A 16 bit word can record 65,536 different values (2^16). At issue is whether 256 different values are enough to convey the range of recordable luminance in each and all channels of color information.

If not, then one can ask if 16 bit word lengths (which is presumed to record a sufficent range) will yield an observable difference in color range and information. There is a fair sized body of knowledge on the science and technology involved, but in the last analysis, it is the quality of the image produced that is the determinant. In short, is there an observable difference, and if so, what is the nature of that difference?

In this investigation, effort is made to make that determination by experimentation with equipment and use (work flow, in the terminology of interest). The standards are those of basic good science: make every effort to determine what is known, what is unknown, and why in each case, such that the investigator is not being fooled by such results as are obtained.

It is a work in progress.

Bill Tallman