"The SD15 got the worst image-quality scores"
The above speaks volumes. The issue is trying to define a subjective issue with objective tools. Image quality is something which isn't directly amenable to numbers.
I agree with the sentiment, Lin, in that 'image quality' is mentioned here so often but I've yet to read a satisfactory definition of it. Anybody got a link?
I doubt that a satisfactory definition will ever be possible because of subjectivity. It's that very issue that goes to the heart of the problem. What makes one oil painting superior to another? Recently a Paul Gauguin painting, "When Will You Marry?" sold for three hundred million dollars!
When Will You Marry? Sold for $300,000,000 This Year!....
In a careful examination of the above painting, what makes it worth $300,000,000 is beyond my ability to comprehend. It it ultra realistic? No. Are the colors different that those in other paintings? No. Is the subject matter something unique? No. So what we are dealing with here IMHO is the fact that there is only one and it was painted by someone who's work someone else who had three hundred million dollars wanted badly enough to part with it. I have a friend who is an artist and in my subjective opinion does far better work than this and feels lucky to get $1500 for one of her superb oils. But then I'm not an art critic although I made my living photographing art for many years.
Unfortunately there exists a sub-set of the human population who persist in believing that virtually everything can be described, sorted and ranked with numbers.
I resemble that remark. Would prefer the term "super-set" though ;-)
Yes, I suppose which side of the gaussian distribution on perhaps that second standard deviation one lies is relevant here - LOL. An interesting example of subjectivism ;-)
To do this, numbers are arbitrarily assigned to various qualities then these numbers are used to make the determination about ranking. After all, it's very easy to rank numbers. So in the end, the "numbers" are correctly ranked but sadly may have absolutely no relationship to the quality under discussion.
I would suggest that is impossible to rank numbers for "various qualities" because "subjective" implies an individual's preferences. In other words, the ranking or weighting itself depends on the individual. For example, I rank sharpness high on the list and I am far less bothered by poor color. Others might prefer good color and not give a toss about sharpness.
You're making my point precisely... Subjectivity simply isn't amenable to assigning numbers because the ranking importance of various qualities themselves are subjective. This means, I suppose, that the qualitative rank of the Sigma was correct for the reviewer but relatively meaningless for many others.
And then there is the big gap between what the sensor saw and someone's final image. In other words individuals' skill comes into play, or even editor performance i.e. ACR's treatment of SD15 X3Fs, as already mentioned. or DP2s or DP2x in FastStone Viewer.
Yes...
Like flower shots: people shoot flowers which often have chromaticities which are outside of human vision but are captured by our fine Foveon sensors. Then people process in ProPhoto because they always do. Even if it looks OK, the temptation to reach out and crank the saturation a bit to make the colors "nicer" can be irresistable. Then they publish to web and, guess what, a horribly over-saturated image appears for our consideration and which many people seem to like, BTW. "Wow, look at those reds!". So, such images are deemed to be of high "color quality" and pleasing to the eye (but best not run a color-picker over the petals to see the numbers, eh?).
[my standard flower shot rant, sorry].
Not a problem! It's always fun to discuss how the human mind and brain react to the visible world and how individuals differ in their perceptions and beliefs about what constitutes "quality." Attempting to rank subjective feelings such as love, hate, desire, etc., becomes rather than a symphony; a cacophony. Are these emotions linear or logarithmic? Which are more important? It is certainly possible to take one hundred people drawn randomly and ask them which of four images they prefer. Then fairly say that perhaps sixty of the one hundred chose image three. In some way then we have assigned a proper number to a subjective term "preference." However, we still haven't a clue which of a multitude of subjective feelings went into the determination by these sixty individuals. Probably even those who chose image three couldn't accurately assess why they liked one more than another. Oh, they could say it was brighter, had more "pop," was more realistic, etc., but the bottom line is there was something which appealed to them and pinning that down to an objective criterion and assigning a corresponding number is where the real issue lies...
Lin
