Originally posted by bossa Famous last words "The dog was watching a horse" Ah yes, that explains everything
This thread is OOC: People seem to be expecting a smaller sensor, of the same generation tech, to outperform a larger one.
There's around one stop difference in noise in these sensors and there's around one stop difference in DOF in the equivalents lenses, so, as I said pages back, why not accept the difference and just try to take the same picture using equivalent lens and ISO settings, handicapping the FF by one stop ISO?
I'm pretty sure the DA*55 @ f/2.8 is a very sharp lens which would give my Sigma 85 F1.4 (@ f/4) on FF a run for it's money where the same number of pixels is involved. In fact, I'll predict that the K-3 would match the D600 IQ at any Hi-ISO in that scenario, where the K-3 was allowed one stop ISO advantage.
Comparing sharpness where both lenses are set to f/4 and where DOF is vastly different due to FL is delusional to say the least.
They expect it, because it happens all the time. They aren't delusional.
And the reason for that is not the science.
Performance isn't measured in numbers, it's measured in how good an image looks. WHen Ilook at my pictures, I can't look at them and say, "this one is 400 ISO, this one is an 800." I have to look at the EXIF.
Years ago, there was the stupid Coke , Pepsi challenge on TV. 50% of coke drinkers prefer Pepsi. You might think that was impressive. Not if you know anything about statistics. 50% of coke drinkers drinking Pepsi means, you can't tell the difference. Now was there a measurable difference. of course there was, scientists could measure it , quantify it, tell you how much more of one substance, how much less than another. That didn't matter, because the people tasting the drinks couldn't tell the difference. They would declare up and down that they were a Coke drinker or Pepsi drinker, but in a blind taste test they couldn't tell the difference.
At this point I can take the images posted on the blog, and show you exactly how that is happening here with these numerical debates. First off, the image are labelled. You know, if you poor Coke into a Pepsi can, Coke drinkers probably won't notice. You get the same effect with labelled images. When I go through the blog posted images I see a saw off. Some people see all the K-3 images being better, some see all the D600 images being better. I see the images being better or worse based on the focus point selected by the AF, not on which camera took the picture. My evaluation of which was better, and I'm saying they're about equal, is based on the whole system. My conclusion is whichever system is better depends on which system nailed the focus, not what sensor was used etc. And the K-3 did quite well at nailing focus, as good as the D600.
So having established that one camera is one stop better at ISO and whatever... what you have to do next is prove that what you have there is a meaningful statistic. Do 1000 humans looking at one picture or another, notice a one stop difference in ISO or whatever. I know I notice the difference between 3 stops difference 100 to 800, maybe 25% of the time.
On some images a 3 stop difference is noticeable, but I'm guessing it's not noticeable ( if you are not comparing the images side by side) in every picture until you get to 4 or 5 stops if you start at base ISO.
So before you can say the one stop difference is meaningful, you have to prove that it's noticeable, and what the conditions are around that. YOU have to prove you have a meaningful difference.
Quote: Comparing sharpness where both lenses are set to f/4 and where DOF is vastly different due to FL is delusional to say the least.
Think of this a taste testing a cup of coffee. A guy says he like two teaspoons of sugar in his coffeee. But you (knowinf sugar is poison) start give him an 8th of a teaspoon less every day, until he says there's not enough sugar in his coffee. So you go back up and ad another 8th, and it turns out I likes his coffee with 1 5/8 cups of coffee. You can say that with this person, one half teaspoon of sugar is a significant difference. That is the level of difference that he can perceptive.
Now think of a quarter teaspoon of sugar as one stop. It may not be enough to be perceived as different by a given taster, but if it is the difference between 1/38 and 1/5/8 where 1.5 teaspoons defines the tasters bliss point it can be huge. So you have the effect of comparing side by side tests, level of sensitivity, and bliss point all contributing factors in how much sugar to put in the guys's coffee. This is just a cup of coffee, things like noise, and DoF are much more complex.
Until you establish that one stop of DoF, or one stop of noise, or one stop of anything is a meaningful statistic, all you've established is that there is a difference.
And that's why all these numerical analysis about a stop of noise, or a stop of whatever are voodoo science. You never established in blind tests with a large population sample, what a significant difference is. ANd while it's possible each of us has a pretty good idea for ourselves what a significant distance is for ourselves, that has no place in a public forum accept as anecdotal information. It could happen that your perception isn't shared by anyone else on the forum, or by everyone else on the forum. You just don't know.
So, for those who persist in this absolutely childish game of numerical differentiation, I can say absolutely, you can talk all you want about numerical differences, but until you've figured out what they mean, you're talking hot air.
The absolute bozo's who started this discussion by posting this comparison actually did a great service, they showed that other factors beside which is an FF and which is APS-c determine how good the picture is, and that case by case, the APS-c holds it's own, in every day shooting circumstances.
I know the lab rats are going to come in here and totally put this down, because the lab rats understand a different language that excludes human differences and perception and focuses on absolutes. And that's valuable stuff... but never depend on them to interpret what their numbers mean to a human being. They are often really bad at that.
So what would I like out of this. I'd like people to stop talking about numerical diffences unless they can show they are meaningful. Useing the coffee thing, are you talking about an 8th of a teaspoon or a 5/8ths of a teaspoon? A difference that's noticeable, or a difference so fine most people can't see any difference. Until you can make that kind of call, you've got nothing.
Oh, and good morning everyone.
Last edited by normhead; 10-28-2013 at 10:50 AM.