Over on Digital Photography Review, a poster whose opinion I respect said something like: “Why are you thinking about buying really expensive lenses; you won’t see any difference unless you’re making five-foot prints.”
I’d heard that before. I’d also heard that you could tell the difference between images from good lenses at screen resolution. Who to believe?
You know me. I had to devise a test. I decided to make prints of the size I most often use: 14×19 inches on C-size (17×22 inch) paper. That gives me an inch and a half border around the image area, which is enough for the framer to use photo corners and still have room for generous mat spacing. I picked a moderately high-resolution camera, the Nikon D810. By full-frame camera standards, it’s very high resolution, with no currently shipping FF camera having more.
I chose what I considered to be a very good lens, and one that I classify as superb. The former is the AF-S Nikon 85mm f/1.4 G, and the latter is the Zeiss 85mm f/1.4 Otus.
Here’s the scene:
I made images with both lenses with the camera on a RRS tripod, Arca Swiss C1 head, electronic first curtain shutter invoked, mirror-up mode, with a 3 second shutter delay. ISO was set to 64. I focused wide open, and made exposures at f/2.8, 4, 5.6, 8, 11 and 16, letting the camera pick the shutter speed.
I brought the images into Lightroom with the default settings and pixel peeped. There were significant differences at f/2.8, moderate ones at f/4, and small ones at f/5.6. After that, the images appeared to be virtual clones.
Here’s a pair of f/2.8 crops enlarged 2:1. On the left is the Nikon lens, and the Zeiss is on the right. Click on the image and view at 1:1 to see what’s going on.
Quite a difference in pixel-peeping mode, isn’t there?
I printed the f/2.8 and f/5.6 images on C-sized Epson Exhibition Fiber paper, using an Epson 4900 printer. I looked. The differences that I’d seen at f/2.8 didn’t seem so important. In fact, they weren’t all that easy to see. I scanned the prints with an Epson V900 scanner at 720 ppi.
The f/2.8 images, with the Zeiss on the top and the Nikon on the bottom:
Again, click on the image and adjust your browser to show it at 1:1.
The differences that we saw in the Lr screen shot are still there, but diminished considerably.
The Otus has greater contrast, as it suffers less from veiling flare. Except for that and the color balance of each lens, there’s no material difference. This blowup is a representation of at least as close as the closest you could get to a print without a loupe.
So, at the kind of apertures that I would normally use, I think that my twist on what the DPR poster said was right: the difference between a very good and a truly excellent lens is not at all important in a C-size print.
What about lens character and drawing? I’ll have to think on that.
I’ve been asked for some screen shots of the area in the image for which I posted the scans of the prints. Here you go. In all cases, the Nikon lens is on the left, and the Zeiss on the right. Magnification is 2:1.
Note how different the scale is from the print scans. That means when we’re pixel peeping at 2:1, we are looking at details that we’d only see with a loupe or in massive prints.
As a bonus, here’s an area with a little more texture, and a very fine wire.
I totally agree with you, Jim.
Sure puts things in perspective!
I’ll be interested to see how you go about evaluating rendering characteristics. As you certainly know, it’s such a subjective range of qualities. Leads nicely into your next post about photo forums…. I’m thinking about the many discussions about 3D-ness and how to achieve it!
we need more of those better rendering lenses than super duper sharp lenses, and that’s why I think Nikon’s design for 58mm f/1.4 is a nice compromise
Sharpness is not everything, but beside being devilishly sharp, the Otus shows the look of a zeiss lens. I have a print destined for a gallery from a scene shot with an A7II and a Leica 28mm Elmarit R, and also the same scene shot with an A7R and the Sony/Zeiss 16-35, also at 28mm. The difference is discernable, with the Leica being the sharper of the two, AND giving more contrast. I haven’t decided which one I like best.
This is really interesting. So if the largest we are likely to print is 19×14, then there seems little point in spending £3,500 on a lens, when the output from the £1000 lens is indistinguishable. For an amateur though, £1000 is still a lot – what about a £400 lens or even a zoom. Would it be possible to repeat the experiment with an even more affordable lens in the mix? And if this is clearly distinguishable at 19×14, then what about at half this size (typical size for most home printers). Thanks!