Note to Mike. The prints sizes that I refer to are based on the standard viewing distance for a rectangular image. Just checked Google again and that is 1.5 to 2 times the diagonal of the "print". Doing the 345 rule for a rectangle this means the diagonal for a 36 x 48 inch print is 60 inches. So the typical viewing distance is 90 to 120 inches or 7.5 to 10 feet. At these distances the visual acuity of 20/20 Human vision is 75 to 56 line pairs per mm. Basically Human Vision has rather poor "resolving power". It's also why the pixel peeping we all do is more about Stroking our Ego rather than if a lens is "good Enough".
This is actually a subject that is more complex than I have time for at the moment, but I wanted to tackle this a bit now.
First off, I understand the general gist of your sentiment in that there are far too many people who are zealots of chart tests, but at the same time, saying pixel peeping is stroking our ego is frankly insulting to a lot of people and doesn't really represent what is going on.
I'd also want to say that the viewing distance of a print is actually a bit *closer* when a good photographic print is in play. It's not uncommon for me to look at a print at 1.25X diagonal, but it varies, and that's not really enough to change a lot of this anyway.
Point #1 of course is that we all have different standards for image quality. All of us. That runs the gamut from folks who think a 20"x30" poster print from Kodacolor X is good enough to those of us (not me) who need 150mp Phase One systems for their work. One thing that is too prevalent in forum discussions is the pushing of everything to the binary side, the poles, where it's either "this" or "that", when reality is there are gradations, and image quality absolutely one of those..
What I generally try to do here is explain the "why" a lot of the time - why a chart test number isn't the final answer to a lenses optical performance, and in this post, why I believe you are incorrect with your print assertion you so commonly make. While I am deeply educated in the technical of photography, I also test for a living, and one thing I often do is take a look at a myth, concept, saying, or supposed truism and test it out. I'm not retired, so I don't have the time someone like Jim Kasson does, but it's something I like to do. And then I try to work through the "why is this so" when what I find might fly counter to the supposed truism, and in this post - your thoughts about lenses needing only to be good enough to produce a large print - I will do so.
Back when the D800E came out, while in the process of long term evaluation of my lenses since I wasn't happy with the output from many of the F mount Nikkors of the time, and this process included looking at third party glass, I ran through extensive/thorough testing - not "I shot this lens in a camera store so now I can talk about it with certainty in a forum", and then made prints. Like you, I believe a medium sized print tells us a lot about a lenses image quality, and if I didn't have competing things for my time, I'd also say I'm trying to work through an exercise of *which* image quality parameters/differences in lenses show up in print versus on a good monitor, but I do have some thoughts based upon experience.
One of the things I did was take two lenses - the Nikon 24/1.4G, and the Sigma 24/1.4 Art, on the D800E, and shot a scene with side to side/front to back detail at F/9 - you know, the aperture where everything should be equal (not) and made 17x22, 16x20 and A2 prints (I tested on different paper types) and then did a BLIND print test. To say that was interesting would be an understatement, because even at SANE viewing distances - about 1.4X diag distance, EVERY participant could EASILY see which print was better. Note that the "loser" was not a "bad" print, but there always was a winner. In other words, my blind print test - which I later repeated with another set of lenses and then once again in the D850 era, showed your assertion to be wrong. Now you can get all worked up about that, but that's that and you can't argue against that. What I think you should do is think precisely what I did, and that is to ask "Why is this so?" because it shocked me.
And that's because of a few things, which I don't really have the time to expound on beyond this post right now:
- Your concept of print size and eye resolution is based IMO on 1970's thinking in the film era. Things are different in the digital era, and there are also other factors that explain why you can see differences at smaller sizes. This is a key point - we are both older, both grew up on film, but I've taken the time to do the research and the tests because I quickly realized a lot of things I knew from RIT in the film era had to be re-evaluated in the digital.
- The primary differences in the lenses that caused this were:
- Better mid/high frequency structure (think in classic optical bench MTF) performance (say the 30-50lp/mm area roughly speaking), which leads to better output. It's not that one print showed more detail, but rather that the overall image looked more realistic. Subjectively, several participants noted the better print "looked livelier, and more realistic". Very long ago I remember a discussion between noted dye transfer printer/tech expert "Ctien" and Michael Johnson, back when they got along, where they talked about the visual impact of better high frequency performance (assuming of course, the subject had such fine detail and it was captured), and IIRC, the sense of reality and liveliness was mentioned. Anyone who remembers this better is free to chime in and correct me as it's been a long time.
- More even performance across the frame. One thing I've found repeatedly in my print-based lens evaluation is that the human eye/viewer will always prefer an image from a lens that is more consistent across the frame as opposed to one with peaks and valleys, even if subtle/moderate, from wave field curvature or wave astigmatism. I believe this is so because consistency is natural, while hot/cold is not. So a lens gamed for a test chart to score well in the corners but right next to it a zone that isn't tested and the lens isn't so good - doesn't work out so well in real life.
- Related to the above, I then researched lens performance and found that some aberrations went away as you stopped down, and some didn't, and the troublesome ones for landscape photographers, field curvature and astigmatism, do not. That's why the "every lens is the same at F/8" myth is false.
- Less of the pesky aberrations (see 2.1 above). If you've got astigmatism creeping in, and now you sample it when you capture it, it's not going away. It may or may not "add more detail" to the image, but you certainly will prefer looking at an image that has less of it. Again, the better vs more concept is in play.
- Now we have to talk about the digital world we live in today. I'm not an engineer, and I don't design sensors, but from research as well as looking at what folks like Panavision in the cine world have to say, my *current* thinking is:
- More data is better than less
- Corollary to the above: BETTER data is better to have
- Acquisition quality should exceed exhibition quality, because in a digital system, where we're sampling and reconstructing, higher resolution, less lens artifacts, less "problems" equals a more natural result. Again, note that this is not about being able to necessarily see more detail, but that the overall image is better. More and Better are different. And it's this concept that is ignored in your more simplistic print size/eye resolution paradigm. Why did an 11x14" print I saw in 1980 from a Hasselbad look better than mine from a Nikon? Surely the increased resolution mattered some, but that wasn't the full tale, and now that we have sampling and reconstruction and so forth, plus we want the ability to perform operations in post on an image, where more/better information is desirable, more/better is certainly valuable.
- Interestingly, there is an older Panavision video from when they announced the 8K DXL or something cine camera, and they showed where even when the output was downsized to a "small" resolution (a 2K or even HD screen), the image from the higher resolution camera was noticeably better. Again, it's a different world than when we were squarely in the film days.
- The complicated part comes next. There is a thought pattern and research being done on what's called hyper-acuity. While human vision is typically thought to only resolve 1 arc/minute (which I believe is like 5-7 lp/mm at 20", not your numbers, but if someone has the math to correct me, I'm open to it), the brain has additional processing that "knows" shapes/patterns/tones, and thus infers things. This was also references in the old Panavision video, so there are odds that there is more than just core resolution going on - the human vision system is both basic and complex at the same time.
At the end of the day, this is why I strongly disagree with your particular print size assertions. In my testing, for my standards, at 13x19 you can barely see differences, but at 16x20 you certainly can. Your standards might be different, and other readers again different. But they are there, and I do firmly believe most can tell, for the reasons I put forth. It's what the blind tests I've done showed too.
Where we likely AGREE is that there is a definite point when good enough is good enough. For the OP of this thread, it would be pointless to get a 45mp camera or even 8K video camera; it's overkill because there is a zone of efficiency, for lack of better wording, between the maximum achievable and what's good enough. And we agree prints are excellent to judge image quality, although I add that careful evaluation of on screen images augmented by printing is the best way. And I'm sure we agree that too much focus on test chart numbers isn't useful.
Respectfully, from another old-time
-mike
edit: PS, my current thinking is that 36mp "got us there" with excellent lenses and print sizes that weren't large format gallery sizes, and 45/50mp gives us some headroom to crop for compositional reasons and a bit of "leeway" in post processing. However, I met a guy pre pandemic who made 12 foot prints and needed everything the phase one could give him; his standards were higher. But I've been at this for a while, and I think for the shooter who has standards that merit a "high" to "very high" level, 36mp to 45mp is a very nice place to be. But that doesn't mean I'm going to knock the guys who moved to the Fuji or HB 100mp medium format backs. But at some point, the gains are getting mini incremental for all but the largest work. IMO anyway.