Discussion:
Apply transmission / absorption spectrum to image
Martin Lederhilger
2014-09-01 16:41:07 UTC
Permalink
Hi,

does anybody know if there is a tool to apply a transmission spectrum
to an image (eg. sRGB -> XYZ -> spectrum -> apply transmission spectrum
-> convert back for each pixel in the image). I would like to calculate
colour correctly, how an image would look like, if looked through float
glass (green).

Thank you for your answers,

Martin
Richard Kirk
2014-09-01 19:58:33 UTC
Permalink
Post by Martin Lederhilger
does anybody know if there is a tool to apply a transmission spectrum
to an image (eg. sRGB -> XYZ -> spectrum -> apply transmission spectrum
-> convert back for each pixel in the image). I would like to calculate
colour correctly, how an image would look like, if looked through float
glass (green).
Hi, Martin.

I have done a lot of this sort of thing. There is not a unique answer, and then you can get some unexpected effects: if you look at tungsten light through cobalt blue glass and amber glass, the two combined look red. If you are wanting to see how a broad spectrum scene would look thorough a particular green glass. The best I have done is to calculate the smoothest spectra for each real (x,y) colour and transform that. This still will not necessarily fit your particular spectrum, but unless you are using something mad like yttrium glass, most reflection spectra are pretty smooth. I can then generate a transform list from these values, and interpolate a cube from the list values. But I cannot produce a formula.

If you need an approximate formula This is a bit specialized. If you want something that just gives the right effect, you could take your RGB channels and yellow and cyan bands between R&G and between G&B. If you have dense combinations of CMY dyes, the transmission often peaks in these two bands, so they can have a significant effect.

Cheers.
Richard Kirk
---
FilmLight Ltd, Artists House, 14-15 Manette Street, London W1D 4AP
Tel: +44 (0)20 7292 0400 Fax: +44 (0)20 7292 0401

Continue reading on narkive:
Loading...