[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Green Light

How nice to find a place on the Internet with high SNR, where artists, craftsmen, and scientists meet!

The crux of the "green light" issue - whether to have wideband or narrowband optical filters and light sources - depends not upon whether additive or subtractive reproduction is used, but whether an original scene or a previously-recorded image is being captured.

In capturing an original scene, the capture device needs to have sensors (or filters) whose spectral sensitivities are fairly closely related to those of human vision. If you believe in classical color science, then use the x-bar, y-bar, and z-bar color matching functions (CMFs) of the CIE Standard Observer. Their shape is quite complex, but these CMFs need to be fairly wideband. Some mathematical processing - for example, a 3x3 matrix - might be necessary or desirable to optimize between the sensor sensitivities and the desired spectral characteristics.

When capturing - or should I say, "recapturing" - information from a previously-recorded image, your objective is to recover the three "records" made previously - if you like, the three numbers already recorded for each pixel. Here it is important is to avoid spectral overlap: Overlap would be difficult to remove, downstream. So a fairly narrowband response to the medium is best. You might choose wider bands for signal-to-noise reasons. However, if the bands are so wide as to admit any overlap between dyes, then downstream correction will be necessary, and that is likely to be a pain. (Correction of this "crosstalk" will introduce SNR penalties of its own.) 

The response to the film spectral absorbtion is the wavelength-to-wavelength product of:

- the SPD of the light source,

- the spectral absorbtion of each of the three actual filters,

- the spectral sensitivity of the sensor device itself, and

- the spectral absorbtion of anything else in the optical path that
   gets in the way!

In the telecine case, the filters with an intimate relationship to human visual perception (or color science) are built into the emulsion of the film. What you should do at the telecine involves optics, physics and electronics, but not perception: You should simply extract the three records. Narrowband response suffices.

If you were to use wideband CMFs appropriate for an original scene - for example, the idealized CIE curves - to sense film, you would reproduce some approximation of the colors reproduced by the film, rather than some approximation of the colors in the original scene.

[Mandatory compensation disclaimer: In Canada, money comes in colors other than green.]



Charles Poynton 
<mailto:poynton at poynton.com> [Mac Eudora, MIME, BinHex, uu]

Thanks to Peter Stansfield for supporting the TIG in 1998..
No product marketing allowed on the main TIG.  Contact rob at alegria.com
944 subscribers in 36 countries on Tue Mar 24 07:14:31 PST 1998
complete information on the TIG website http://www.alegria.com/tig3/