Peter Swinson on Scanning Colorimetry

From TIGwiki2
Jump to: navigation, search

A few people have asked me about the colormetrics of scanning. As Kevin Wheatley from Cinesite UK pointed out there is always more to life than one expects! Keith made reference to SMPTE RP 180 which defines the taking characteristics for film scanners. Whether they are adhered to or not is another matter.

However I took the SMPTE figures into MS excel and plotted them.

SMPTE RP-180 plot.jpg

I then overlaid them onto a typical Kodak neg stock 5277. I only used the old Vision 5277 curves because Kodak's leaflet shows them more clearly than the Vision 2 leaflets.

Kodak 5277 Neg Plot.jpg

The results may be of interest.

They clearly show how the SMPTE response includes the scan detectors being effected by color dyes other than the dye layer of interest, in the same way that print stock would "see" the negative.


Scanner heads can simulate this in one of two ways. By matching the RP 180, this has the advantage of gathering most light, area under the curve = volume of photons. Or Have narrow band sensors and then electronically simulate the "bleed" of unwanted dye layers into the channels by matrix/masking.

RP-180Sampling5277 annotated.jpg

Example of "cross coupling". Note the sampling gathers parts of other film dyes (shown by shaded sections.) Narrow colormetric scanner channels, shown dotted, would largely avoid this, give greater saturation, but be untrue to the film response, unless matrixing added back the amounts shown shaded.

There are other other devils in the detail, but I guess if they were all revealed, everyone would be building scanners!!!!

It would be interesting to hear from the Viper and other Electronic aqcuistioners out there whether these cameras follow these "filmic" characteristics. I have long believed that the native Viper "greenish" cast is due to this attempt to match negative film dyes. Is it?

Time to descend into the bunker and avoid the flak

Peter Swinson

Original composite image (1134x1325, 90 KB):