Originally posted by Alex645 I think it is very telling that in DXOmark's somewhat outdated sensor tests that MF and FF fill the top 33 spots before the first APS-C:
Camera Database - DxOMark
In a 72 ppi, sRGB color space comparison, the difference is going to be minimal to insignificant as normhead and Jesse have shown us. But at 300-600 dpi, AbobeRGB color space, printed 13x19" or larger, the differences become more apparent, especially in higher contrast scenes, higher ISOs, etc.
As I've said before, ultimately the lens is going to be the biggest asset or liability to the image, but if all things are equal, size matters (yes, yes, with the understanding that the 'focal length is the focal length') not only in the sensor but also in the viewing of the screen or print and the viewer's distance from the media.
Only past the point of oversampling. At 300 DPI, which is still a lot for most purposes, a 19 inch print needs 5700 pixels, so a K-1 and even a K-3 oversample the images. Any body K-3 or higher is an oversample of the required resolution. I've never seen evidence to suggest a higher degree of over sample leads to more resolution or abetter image especially at 100 ISO.
I have 16x20 prints of my window ice crystals shot with my K-3 and my K-1, and I've forgotten which image was taken with which camera and can't tell the difference. So, it is conceivable that at some point it makes a difference, but I'd prefer to here that from someone who has actual prints that show when that difference becomes apparent.
Without such references, and with concepts like an image that is larger is going to be viewed from further away if it's bigger I am reluctant to make such judgements.
Photography tends to be dependant on double and half type differences. A K-5 is 5000 pixels wide a K-1 is 7400 pixels wide. Folks will argue that makes a difference, yet on my biggest viewing surface. my 4k 55" TV, a K-1 and K-5 image are virtually identical. On a 4k TV which has become my favourite method of looking at large "prints" even a K-5 is an oversample.
So, in short, follow such logic at your own peril. You may not get what you think you are going to get. And I'm not going to believe it's true until I see some definitive proof, not pixel peeping, but evaluating the whole printed image. I already know it makes no difference on my 4k TV. And I also know it makes difference for the largest prints I do.
And with modern vector based enlarging software, enlarging an image can often clean up the mess left by too much resolution and the capture of irrelevant and messy looking detail.
Most of us don't care if there's a little more detail on dogs wisker. It amay be there, but the difference is not within the range of human perception. I don't examine prints with a loupe.
Of all the people who've claimed you'd see a better result at larger sizes no one has ever produced a print comparison to show that's true. The only person who tried, compared a K-5 to D800, and in blind test, the test participant couldn't makeup their mind which they preferred. That's the only evidence I have to date, and it doesn't point to a resolution increase of even 100% leading to increased enjoyment of a print.
I've waited over 5 years now for the counter point, and it simply hasn't been forthcoming. At some point (like 5 years ago for me) you have to decide it isn't there because no one has done it, leading us to suspect it can't be done. Yet people go on as if this were a proven fact.
What it is is commonly stated unproven assumption that through repetition has come to be accepted as fact, with absolutely no evidence beyond people saying its what "should" be true.