This is the 41st in a series of posts on the Fujifilm GFX-50S. The series starts here.
I did a series of informal tests comparing the Sony a7RII with the Otus 55 against the Fujifilm GFX 50s and the Otus 85. In highly magnified images, the GFX had better resolution and quality.
But we don’t take pictures to look at tiny sections on computer screens, do we? So, the question on the floor today is how large do you have to print to see much difference?
Here’s the way that the GFX looked at the scene:
Using an Epson 4900 and C-size Exhibition Fiber paper, I printed out square crops using the full image height 15 inches high from both cameras at f/5.6 for the a7RII and f/8 for the GFX, to equalize depth of field. That gave the a7RII a small advantage, since both lenses are sharper at f/5.6 than they are at f/8.
I couldn’t see any material difference.
Then I cropped both images to half their height and made a pair of 15-inch-high prints. That’s what you’d see if you looked at a section of a 30-inch-high print of the full height of the image.
Then there was a difference, in favor of the GFX. Not a huge one, but certainly one that could be of significance to perfectionistic photographers printing that big.
Next, I looked at the f/4 image from the a7RII and the f/4 image from the GFX. There was still virtually no difference between the resolution of the 15-inch-high prints, and the difference between the 30-inch-high ones was larger, although well short of striking.
When I looked at the two f/2.8 images at 30-inch print heights, there was quite a bit of difference towards the edges in favor of the GFX. I’m interpreting this to mean at wide apertures, the advantage gained in the case of the a7RII by not using the very periphery of the image circle was less than that of the GFX ‘s edge in having more coarse sensor pitch and greater resolution. In this case, you can see some differences in even a 15-inch-high print, but you’d never notice it if you weren’t looking for it and had the two shots side by side.
A couple of observations:
- The most important difference between the prints wasn’t that the GFX ones looked sharper, but that they looked somehow more real.
- Inkjet prints are a long, long way from matching the resolution of a silver gelatin contact print or a photographic transparency.
So now you know. If you’re a stickler for quality and print 30x40s, you’re going to like the GFX better than the a7RII, not even considering the relative quality of the native lenses. If you have a 13 inch printer or print C-size and you’re going to shoot with the very best (mostly adapted) lenses, you might as well save your money and buy an a7RII. But this was a sensor test, and you should consider that the native lenses for the a7RII are, as a group, a considerable step down from the native GFX lenses.
Thanks Jim!
This is the kind of thing we need more of (and haven’t seen since pop photos SQF idea which was always a bit vague) – an effort to tell if and under what conditions technical differences are visible .
At what distance did you examine the prints, btw?
Between a foot and eight inches.
Actually, Ed Grainger’s (sp?) SQF metric was quite specific. Dunno if Pop Photo used all of his methodology, though.
Yes I don’t recall if they did – or if the did whether they regularly explained it.
So your tests are for close inspection. Tempted to try a test which determines what size print is the cutoff for visibility when viewed at the closest distance you can comfortably scan the whole print with your eyes…
If I kept backing up so that the entire print could be comfortably observed as a whole, my eyes couldn’t see the difference no matter how big the print was. Maybe younger eyes, though.
Hi Jim,
A few comments:
Ed Granger’s SQF is probably well accepted as a relevant metric of perceived image quality. In essence it is just the integral over the product of MTF of the system multiplied by the Contrast Sensivity Funtion of the eye.
One aspect is with SQF is that it is heavily dependent on sharpening. With a certain amount of sharpening it is possible to get very good SQF.
So, in a real world sharpening plays a role. All images need proper sharpening, an I think we all agree on that.
What proper sharpening is can be discussed. I would guess that my ideas about correct sharpening may differ from Jim’s. You know, Jim is the master and I am the batchelor – listening to the master and learning a lot.
The way things are, the better gear you have the less sharpening should you do. Sharpening cannot replace resolution but can do a decent job of compensating for loss contrast.
Back in 2008 when I graduated from 12 MP APS-C to 24 MP on full frame I was astounded to see how good A2 (C-size) prints could be made from 12 MP APS-C.
Going from 24MP on 24×36 to 39 MP on 37×49 MFD I could not really observe benefits of the 39 MFD over the 24 MP FF in A2-size prints, without a loupe, that is. With a loupe the benefits of the higher resolution were quiet obvious.
Interestingly enough, SQF as calculated by Imatest confirmed what I could see:
http://echophoto.dnsalias.net/ekr/index.php/photoarticles/82-why-i-cannot-see-a-difference-in-a2-size-prints
Especially, using Imatest data with “standard sharpening” all systems I compared yielded 98-99 SQF for A2-print size.
No, Imatest “standard sharpening” is not a recommended norm. It is intended as a tool to compare sharpened images. What it suggests is to aim for a sharpening that “lifts” MTF to 100% at low and medium frequencies avoiding oversharpening low frequencies and avoid sharpening detail near Nyquist. It may be a decent recipe…
Best regards
Erik
“Inket prints are a long long way from matching the resolution of 8…) a photographic tranparency.”
And what should we make from this insight?
Whenever I do show tranparencies, poeple poke a little bit fun of me for being so outdated, but they usually quite like, what they see.
But as far as prints are concerned?
(I’ve switched to the a7r with Loxias mainly because of convenience).
I think that there is the possibility of higher-resolution digital output media. In fact, there are photographic output processes that (used to, at least) deliver far greater resolution on a pixels/mm basis than we have now with inkjet. When those processes come along, we will have to revise our estimatation of what file resolution is OK for what size print.
If you’ve ever held an 8×10 chrome — the very film that was in the camera at capture — you know what I’m talking bout. 8×10 and 8×20 silver contact prints come close to that.
Jim
Currently i’m toying with upgrading my A7r2 to the Fuji GFX. Tried out a 150 f4 V series Hasselblad lens today on the Fuji GFX with an adapter and appears very good to excellent when stopped down to f8-11. Its not as sharp as the 120 fuji that i also tried but with some work in Iridient it comes close enough to (in my opinion) not matter. Actually I’m finding Iridient gives cleaner sharper files than Lightroom with all my cameras. I also have a few Mamiya 645 lenses (55 2.8, 80 2.8 and 300 5.6) that i acquired very cheaply that could be adapted. Not sure how well they will work but I understand that the 55 & 80 at least are considered to be sharp lenses.
However the thing that bothers me about upgrading is the thought the Sony will come out with a much better A7rll in the next year or so. I’m fairly impressed with my A7rll, but don’t think its got there yet – re sharpness for fine detail in landscape, hence the interest in the Fuji. Maybe the combination of increased resolution, dynamic range and reduced noise in the Fuji is more important than resolution increases alone?
Printing Sizes.
I’ve never used 8×10 cameras but had a 4×5 field camera for a few years. The transparencies, especially Velvia, were very impressive on a light table with loop. Re Print sizes – What did surprise me many many years ago was receiving prints from my first medium format camera an old 6×6 120 film folding Zeiss ikon. the prints were only 4×4 inch but had beautiful tonal graduations and appeared very sharp at that size. I thought at the time how much better they looked than prints from my 35mm camera. Possibly the quality difference between medium format film (645, 6×6 and 6×7) and 35mm was much greater than for corresponding digital sensor size differences when printed?. For me the ‘medium format look’ is smooth tones, no grain and overall crunchy sharpness.
I recall that prints (even only 8×10 ) off my 67 Pentax were much sharper and with better tonal smoothness (lack of grain) than anything I got off 35mm. More recently I saw some smallish prints off a Sony A7s and they reminded me of medium format film prints. I can understand why people used to make contact prints off 8×10 film.
Thanks for all the detailed work that you do Jim. Way beyond the capability of most of us.