"Gamma error in picture scaling"?

[Dieter Mayr]

Maciej, thank you very much for your explanation, it makes perfectly sense to me.
Doing some tests with "real" pictures i just can agree with Den that it may have not much relevance with those pictures.
Nevertheles it will be good to keep that in mind if a image should make problems when resizing.

[MarkT]

Well, thanks for all the explanations and illustrations. I completely agree that the chance of noticing any irregularities in actual photos seems small. I guess I'm curious to hear from Jonathan and Kiril as to the ROI of implementing a "fix", or if there really isn't any need to worry about it.

Interesting stuff.

[tomczak]

I'm my attempt to explain what Gamma Adjust does, I made at least one error: if Input Gamma is set to your actual display gamma (e.g. 2.22), the Output Gamma is not the gamma to which a new display should be set so that the Output Image looks the same. Rather, to look the same as Input Image, the Output Image should be displayed on a monitor with gamma=(Input Gamma)/(Gamma Factor). The rest of the concepts I probably got not far from truth.

Gamma is a simple concept, but I get it wrong all the time, probably because it's applied in a few places, sometimes explicitly, sometimes implicitly. While Gamma Adjust seems to be easy to explain mathematically (it just raises pixel values to the power=Gamma Factor, while showing, for convenience, the result: Output Gamma=Chosen Input Gamma * Gamma Factor), it may be confusing.

For instance, if my my display is calibrated to gamma 2.22, the image looks good on it, and I want to processed it so that it looks just as good on another display or printing press with gamma 1.8 (which is the 'labeled' use of the transformation), what do I do? Since display gamma 1.8 will not darken the image as much as display gamma 2.22, the image needs to be pre-darkened. The easiest way to do that, set Input Gamma to the desired value of 1.8 and adjust the Gamma Factor slider until the Output Gamma is 2.22. The mathematics work, but I find it counter-intuitive. Perhaps different labeling could help - though I don't know how to do that...

[Dieter Mayr]

Maciej wrote: "The mathematics work, but I find it counter-intuitive. Perhaps different labeling could help - though I don't know how to do that..."
Thats exactly the point that confused me too.

[MikeG]

For what it's worth, here are the the originals plus a 50% resize by cPicture 2.5

Mike.