miguel_13 wrote:
This has been the most enlightening thread I've ever read on this forum.
The sample base isn't really big enough for anything 'publishable' but it l admit I love some of the analysis behind participant's choices.
Given that 90% failed to get it entirely right and 51% got it entirely wrong ... even if you apply confidence factor relative to the sample size ... the story doesn't change much. Would it be a different story for a different scene ...
With 70 of us "casual observers" as the sample, and the results ... I think it is pretty telling (for this scene). Of course, once you make the physical apertures their correspondingly different sizes (i.e. same f-stop) it becomes a different story that would likely yield a very different result.
Jman13 wrote: I did a few shots at 85mm equivalent with the CV 42.5/0.95 on m4/3, the Hexanon 57/1.2 on APS-C and the Rokinon 85/1.4 at f/2 on FF, but it became painfully obvious since the Rokinon not only has exceptionally smooth bokeh at that range (much smoother than the other two) and at f/2, the Rokinon is far sharper than either of the two manual lenses....it became much more a lens test than how the format themselves render.
this is really the major difference between the formats at this point, it's a lot easier to design a narrow dof lens for larger formats. getting extremely well corrected narrow dof is just easier (and cheaper on FF). otherwise there isn't a lot to care about other than super high iso performance.
RustyBug wrote:
Of course, once you make the physical apertures their correspondingly different sizes (i.e. same f-stop) it becomes a different story that would likely yield a very different result.
Meaning I suppose that the optics were manipulated to 'fool you' into not being able to identify the different sensors?
Even more fun would be to use a true random sample of subjects and see if they did any better than the experts in a photography forum.
"Even more fun would be to use a true random sample of subjects and see if they did any better than the experts in a photography forum."
I wonder if us "experts in a photography forum" may "overthink" this sort of comparison, and bias the results - a random selection of folks off the street would likely be closer to pure random results. So, still pretty bad, but better than us...
Good to see I got it right, since I was fairly confident. My base feeling/reasoning was a mixture of:
Jochenb - Pana25 CA
wayne seltzer - by rendering style
And as I said, I was surprised about how much more noise the Canon FF camera had. I've only compared the OM-D with the D800 and D7000 before. But even so, #2 had the most relaxed rendering, the least vignette and the least amount of CA. And the best small-detail contrast, and the smoothest transition in/out of focus. If you had had more light so you could shoot it at longer shutter speeds, it would have been a very easy pick, even at 1600px image size. And you could have shot it with the 120USD 50/1.8 and it still would have bested the fairly expensive Pana-Leica 25/1.4.
BTW, Lightroom does NOT in any way guarantee equal image WB by setting equal WB "temperature" in the controls. How well the set temperature corresponds to reality depends on the camera profile and the light circumstances.
And again, thanks for doing the work with setting up the comparison. This is actually the only way to do it fairly, and any criticism of "fairness" with regards to execution is IMO misplaced since all the necessary information was presented "as is" in the presentation text.
byebyefilm wrote:
Ok, now I'd like to know up to what size can the mft and aps-c shots be enlarged and printed and still look almost identical to the FF one?
based on the noise levels for this particular older generation FF camera in this shot, the aps-c and µ4/3 cameras will look better than the FF at larger sizes.
theSuede wrote:
But even so, #2 had the most relaxed rendering, the least vignette and the least amount of CA. And the best small-detail contrast, and the smoothest transition in/out of focus.
to be fair it also has the greatest dof by a fairly large margin which helps in many of these points.
sebboh wrote: this is really the major difference between the formats at this point, it's a lot easier to design a narrow dof lens for larger formats. getting extremely well corrected narrow dof is just easier (and cheaper on FF). otherwise there isn't a lot to care about other than super high iso performance.
Really? Admittedly, last time I tested m4/3 vs. FF it was GH2 vs. 5DC, and the detail level in FF was what I cared about for landscapey tripod shooting. And FF was much better for that...
Has the gap closed to where fine detail level is nearing the same between m4/3 and FF?
I would be most interested to see this repeated in a landscape setting, with lots of detail (grass, tree leaves, etc.). If I had a m4/3 rig, I'd shoot a set and see for myself.
mco_970 wrote:
Really? Admittedly, last time I tested m4/3 vs. FF it was GH2 vs. 5DC, and the detail level in FF was what I cared about for landscapey tripod shooting. And FF was much better for that...
Has the gap closed to where fine detail level is nearing the same between m4/3 and FF?
I would be most interested to see this repeated in a landscape setting, with lots of detail (grass, tree leaves, etc.). If I had a m4/3 rig, I'd shoot a set and see for myself.
For portrait, the less fine detail the better.
really, you found the 5Dc gave you better detail for landscape than the gh2? what lenses were you using?
sebboh wrote:
really, you found the 5Dc gave you better detail for landscape than the gh2? what lenses were you using?
GH2 had PanaLeica 45 macro on it (which was supposed to be a really good m4/3 lens at the time, no idea how it measures up now), and I think 5D had 35-70 Contax on it. I was trying to figure out if I should carry GH2 and a small prime kit, but it seemed like 5D was much better with a zoom. Granted, 35-70 is not your average zoom...
'but it is interesting to see that for a typical shot'
Focus distance is much closer here and DoF much thinner than a 'typical' image, I'd have thought.
Rather than attempt picks where many factors enter the picture, I like to look at what I prefer in these kinds of comparos then work out why...and #2 is a clear winner because of the little things, the tone range of the more flaxen hair, even the blotchiness on her arm.
These are due to higher and smoother colour contrast (a characteristic Mike Johnston calls 'microcolor' - the variation of colour with contrast) and as a result, a greater sense of reality due to the perception of better shaping, less image 'flatness'. #2 has greater tonal sublety also. How much is lens, how much is format I don't know, have never used any of these.
And yes, it is fun and thanks very much for doing it.
Philip - sorry, I don't see it. Personally, I think the tonal transition is actually worst on the Canon image. The Fuji has the smoothest tonality to my eye, but all are quite close
I find it interesting how people are generally able to see a much bigger difference AFTER the images are revealed for what took what.
You may have a preference for #2, but I don't think it can be said to be clear since the statistics of group observation has shown it's anything but.
mco_970 wrote:
GH2 had PanaLeica 45 macro on it (which was supposed to be a really good m4/3 lens at the time, no idea how it measures up now), and I think 5D had 35-70 Contax on it. I was trying to figure out if I should carry GH2 and a small prime kit, but it seemed like 5D was much better with a zoom. Granted, 35-70 is not your average zoom...
hmm, for landscape i think i would prefer the contax on the µ4/3 camera to the that panaleica (you could have shot it at 35mm on µ4/3 and 70mm on the 5Dc). if you had compared the canon 85/1.8 to the olympus 45/1.8 i think it would be pretty close. also, what aperture? after f/5.6 you can visibly see the pixels get softer due to diffraction on the gh2 if i recall correctly.
with modern sensors you have a the advantage of a lot more pixels with FF, but those are invisible at most common viewing sizes/distances while differences between an f/.95 on lens wide open on µ4/3 and an f/2 lens wide open on FF are obvious at standard web sizes and 3x5 prints.
We'll, I wouldn't go that far. Just that the Rokinon stopped down a stop vs an f/0.95 wide open happens to be easy to spot in portrait processing up close. The 42.5/0.95 is pretty sharp wide open, it's just got a lot of SA at closer forces distances. That goes away pretty quickly. The outdoor shots were easy to spot also due to smoother bokeh from the Rok.
But the 42/0.95 wouldn't be easy to pick out of a lineup on, say this image, taken wide open.
Jman13 wrote:
We'll, I wouldn't go that far. Just that the Rokinon stopped down a stop vs an f/0.95 wide open happens to be easy to spot in portrait processing up close. The 42.5/0.95 is pretty sharp wide open, it's just got a lot of SA at closer forces distances. That goes away pretty quickly. The outdoor shots were easy to spot also due to smoother bokeh from the Rok.
But the 42/0.95 wouldn't be easy to pick out of a lineup on, say this image, taken wide open.
true that image might be difficult (i feel like the rokinon would have noticeably more microcontrast, but maybe not), but all the other f/.95 images you posted on this thread: https://www.fredmiranda.com/forum/topic/1240105/ would not.
When we've had Zeiss vs Non-Zeiss threads in the past, the differences between the photos were generally much more significant and obvious to me than the differences between the photos Jordan posted. Glass definitely matters.
If I had to choose, I'd say I prefer the OM-D shot from the three at the top of this thread, btw. I assume it was shot with the PL 25, and I just enjoy the way that lens renders.
I did a few shots at 85mm equivalent with the CV 42.5/0.95 on m4/3, the Hexanon 57/1.2 on APS-C and the Rokinon 85/1.4 at f/2 on FF, but it became painfully obvious since the Rokinon not only has exceptionally smooth bokeh at that range (much smoother than the other two) and at f/2, the Rokinon is far sharper than either of the two manual lenses....it became much more a lens test than how the format themselves render. 