Negative film - where's the shoulder data?

[Greg Campbell]

I've browsed a number of forums, attempting to determine the 'true' dynamic range of Ektar100 and several other print films. After wading through way too many arguments, I 'noticed' the obvious fact that, as with most negative emulsions, Kodak does not define the shoulder response area of the film.

Looking at this plot, it appears that the red response is just starting to round off, but blue and green density is still increasing at a roughly linear rate.
If we ignore high end color shifts as the channels begin to roll off, just how much raw dynamic range is 'up there'?

Thanks

 

[hrst]

That's true, the plot is NEVER done much over that point, but there definitely IS a shoulder.

I found this test years ago:
http://archive.bigben.id.au/tutorials/360/technical/hdri/index.html

It reveals that there is MUCH data where the normal plotting of curves end.

The shoulder is almost as long as the linear part. So, if we have 9 stops of linear, then there's almost 9 stops of shoulder, and then it's completely blocked. However, only maybe 3-4 stops of this shoulder is usable at all. So, the "real" dynamic range of color negative films is about 14 stops.

I've attached a picture overexposed 9 stops (Fuji Superia X-TRA 400), scanned with Nikon Coolscan V and adjusted (huge contrast increase). Unadjusted image is almost completely white. The whole image is on the shoulder.

 

[Greg Campbell]

Excellent, thank you!

 

[hrst]

If you have to compare with digital, all the digicam manufacturers and fanboys always use the highest possible value for dynamic range, even if it's completely unusable due to noise. (These values range from 11 to 13 stops, real values being about 2-3 stops less.) So, if you deal with digifanboys, you can say that film has a dynamic range of more than 20 stops, because that's completely true on the terms they have set.

 

[Greg Campbell]

If you have to compare with digital, all the digicam manufacturers and fanboys always use the highest possible value for dynamic range, even if it's completely unusable due to noise. (These values range from 11 to 13 stops, real values being about 2-3 stops less.) So, if you deal with digifanboys, you can say that film has a dynamic range of more than 20 stops, because that's completely true on the terms they have set.

I'm sensing you've heard "What's wrong with yout? Haven't you gone digital yet?! :rolleyes:"
a few times too many?

That is a good point, comparing the long, muddy digital toe vs. film's long, dense shoulder. I've heard the digi-boppers make claims that my dslr can't begin to live up to. In the real world I get a hard ~ +2.8, and can squeeze -5, at best, before the s/n starts to get silly.

Thanks again for that link. The author's 'Relative EV vs. Difference in pixel values' graph makes the point very nicely.



All this begs the question: Why do negative film manufacturers terminate the density curve 'prematurely?' All the positive film curves I can remember seeing include both shoulder AND the dense toe region.

 

[David A. Goldfarb]

Transparency film has less dynamic range than negative film, so the toe and shoulder are visible within the normal exposure range included in the published curve. I suppose the manufacturers don't publish the extended curve of negative film, because only so much of the curve is going to be printable anyway.