[HDR-photo] Real-World example (Was: Alignment)

[Bernhard Vogl]

Hello Pablo,

> > - The HDR is technically not correct (too low DR) because the images
> > were merged with a standard response curve. The images were captured
> > with a Fuji S3 in extended range, but i can't use response curve
> > calculation as this would mis-align the HDR brightness of the sectors
> > (...the same problem as above).
> 
> I'm not sure what you mean by this? Is this a limitation of the software
> used for HDR merging, or cause by a variable response curve?

The information about what's exactly happening when using the extended range is not public available, so all i can do is some guessing:
The S3 sensor incorporates 2 types of photosites, S and R pixels. They are roughly 2 f-stops apart. So, one image is the same like a 0EV/-2EV bracketed shot with a normal camera.
Both brackets are merged on the fly at the time of shooting, with also a basic sort of tonemapping happening (depending on the camera settings).

The resulting JPEGs of course can't show this - compared to to a "standard" digital camera - extended range. All you can see is a smoother s-shaped response curve. See http://dativ.at/temp/chart.jpg
(Nikon D70, Fuji S2, Fuji S3 against white screen, bracketed images until pixels "burn out")

So any software that will merge the JPEGs to an HDR (with standard response curve applied) will not account for the extra EV steps...and if you let the software calculate the response curve, the absolute brightness value of the HDR sector shots won't match up any more. To my understanding, i'm missing at least the extra 2 EV steps of dynamic range in the resulting HDR.

I hope i was able to communicate my thoughts...  ;-)

Best regards
Bernhard