D70 Histogram Investigation
Prepared 2004-12-07 (150/121247) by Bill Claff
I performed two
"scientific" experiments regarding the D70 histogram (they would have
been more scientific had I used a tripod).
In discussing this topic I'll be using percentages from 0% to 100% (or numbers from 0 to 1) to describe D70 histogram positions.
I took four photographs of solid white, red, green, and blue images displayed on an LCD panel. The white image was taken first on Auto. The red, green, and blue images were taken on Manual with the same aperture and shutter speed as the white (and focal length). The distance to the image was approximately constant (although I now think the white image was slightly further away).
Each image showed a D70 histogram with a sharp peak. (W=75%, R=25%, G=50%, B=18%)
Each D70 histogram matched the luminosity histogram in Adobe Photoshop CS for that image.
(Note that luminosity is NOT the NTSC formula for luma quoted elsewhere as variations of L = .30*R + .59*G + .11*B)
1) The D70 histogram does not use only green. It shows red, green, and blue information.
2) The D70 histogram is achromatic; it displays luminosity.
The D70 is like the human eye in that it is less sensitive to red and blue than it is to green.
If you're using the D70 histogram to determine exposure and your subject is primarily red or blue then you should be aiming for peaks that are much further left, about 18-25% rather than 50-75%.
I took successive photographs of two targets, a small dark red bar on a bright red background, and a small dark blue bar on a bright blue background, at longer and longer exposures until two of the three channels were totally clipped.
The D70 histograms moved progressively to the right but even 6 stops overexposed they were still visible.
The clipped values did not show on the D70 histograms.
1) The D70 histogram uses CCD values 0 to 4094 but not 4095. Clipped values do not show.
The D70 histogram is not a good indicator of clipping. Use the Highlights display instead.