Lens pr0n - 50s on R-D 1 - part I

[jlw]

It's great to have a day off work with nothing to do until evening! Still having some time in hand before I go off to shoot rehearsal photos, and already having the tripod and my little test scene (from the 10mp-vs-6mp thread) still set up, I decided to stick all my 50mm lenses on my R-D 1 and shoot it to see if I could learn anything interesting.

For all shots, I used EI 200, shot in raw mode, and converted at Adobe Camera Raw's default settings (so don't try to read too much into minor color-balance differences.)

I had four lenses handy that will fit on the R-D 1: a modern C-V Nokton 50/1.5; a Canon 50/1.4; a Canon 50/1.2; and a Canon 50/0.95 converted to M-mount. (Anyone who feels I should have included a Leica, Nikkor, Carl Zeiss, or FSU 50, feel free to lend 'em to me for six months or so and I should be able to get around to it...)

I shot each lens at its full aperture, whatever that might be, and again at f/8. At this point somebody's going to say, "No, you should have shot all of them at all the same apertures so we can see how the 50/1.2 does at 1.5 and so forth." But -- that's not the way I personally use my lenses. Typically I'm either shooting in awful light, where I need all the aperture I can get... or in a studio setting with strobes, where I can stop down all I want. So, doing it this way may not be "scientific," but it provides me with more useful learning for how I shoot pictures.


Rather than posting the sample pix from each lens separately, I'm going to attach comparable sections from each to the same message, so you can see how they look side-by-side. For example, here are photos from all four lenses, showing the test scene at f/8. I have reduced all of them down to an RFF-friendly 560 pixels wide, applied a slight (and standardized) amount of Smart Sharpen filter to compensate for the downsampling, and saved as JPEG at the max quality setting:

 

[jlw]

Lens pr0n - 50s on R-D 1 - part II

Lens pr0n - 50s on R-D 1 - part II

Now here's the same full-scene shot, with the lenses at their maximum apertures.

One thing I've learned already: I didn't expect to be able to see ANY differences in the lenses once they'd been sampled down to 560 pixels wide, but some "flavor" DOES in fact seem to be visible! What do you think?

 

[jlw]

Lens pr0n - 50s on R-D 1 - part III

Lens pr0n - 50s on R-D 1 - part III

Now here's a detail from each scene, with the lenses at f/8. No huge surprises here, other perhaps than that they're surprisingly close...

 

[jlw]

Lens pr0n - 50s on R-D 1 - part IV

Lens pr0n - 50s on R-D 1 - part IV

Now here's the same crop at each lens' full aperture. Am I nuts, or does the 50/0.95 at 0.95 look better than the 50/1.2 at 1.2?

 

[jlw]

Lens pr0n - 50s on R-D 1 - part V

Lens pr0n - 50s on R-D 1 - part V

Here's a crop of a different area at f/8. By the way, all these "crop" images are shown pixel-for-pixel at the original file size -- no downsampling.

 

[jlw]

Lens pr0n - 50s on R-D 1 - part VI

Lens pr0n - 50s on R-D 1 - part VI

And here's the same area with the lenses at their maximum apertures. Since there's no print in this area, the quality is more "subjective," and I think the 50/1.2 does better here than in the other crop. But what do you think?

 

[jlw]

Lens pr0n - 50s on R-D 1 - part VII

Lens pr0n - 50s on R-D 1 - part VII

My takeaway:

-- Before I shot these pictures, I had assumed that the modern 50/1.5 Nokton would be the best all-around, the fairly modern 50/1.4 Canon would be second-best, the older and faster Canon 50/1.2 would be a fairly clear-cut third, and the "extremist" 50/0.95 Canon would be clearly the worst in terms of overall picture appearance.

-- But the real result wasn't so cut-and-dried, was it? At their full apertures, one could argue that the 50/0.95 yields more detailed-looking pictures than the 50/1.2 (which may not be getting a fair shake, as its correction may have been optimized for longer distances; still, I gotta shoot 'em like I've got 'em!)

Meanwhile, the Nokton doesn't seem to have as much advantage over the 45-year-old 50/1.4 Canon as you might expect considering their respective ages, the fact that the Nokton benefits from aspherics and multicoating, and the fact that the Canon gets by with only six elements (it's the only 6-element 50/1.4 I know of, in fact.)

-- All the lenses can shoot nice, sharp-looking pictures at f/8, but there are still some small differences in "look" from one to the next.

-- I had assumed that shrinking the images down to 560 pixels wide and applying smart sharpening to them would make them all look just about exactly the same, but that didn't happen either. Some differences in character still are evident even after this much post-processing.

-- Now I see why some people on RFF own so many 50mm lenses!

 

[pfogle]

Well, I started out as a skeptic, thinking the Nokton would slay the Canons, but I have to admit, I like the 1.4 the best... Oh dear, just when I thought I was satisfied!

Anyway, a nice demo, thanks for the trouble 🙂

 

[Trius]

Wow. Amazingly, I like the 0.95 the best, except for higher contrast situations where the 1.2 would be my choice. I'm surprised the Nokton, while not a bad lens, "suffers" so much in this comparison.

Now ... it would be hard to conduct the test, but I'd like to see the Nokton 40s (both versions) thrown into the mix. 😀

 

[jlw]

Very nice, thanks for doing the work and posting the results!

It seems to me that the focal plane of the 1.2 was sightly different than for the .95.

You're welcome; glad you found it interesting. It wouldn't surprise me if focus accuracy were an issue to some extent. I was using dim overcast daylight, which got dimmer later in the day. I did the 50/0.95 first and the 50/1.5 Nokton last, so the 0.95 probably benefitted from the best focusing accuracy.

Also, this particular 50/0.95 has been hand-collimated specifically to my particular R-D 1. The other lenses have been checked for focusing accuracy at one time or another, but haven't been specifically matched. This may have something to do with the variability of results also.

On the RD-1 the 50 is effectively a 75; even the toughest Leicaphiles would hesitate to focus a 75/0.95 even on an M3 ...

I did use a 1.3x eyepiece magnifier, which helps quite a bit.

 

[jlw]

Is there a way you can check with your RD1 how much faster the
1.4 is compared to the Nokton (if at all) ? The Nokton is effectively rated
f1.6, but I could never find any info on the Canon 1.4.

I tried a couple of things, but I'm not sure what might be a defensible methodology.

Anything involving trying to read out the differences via the camera's metering system is going to run into problems because the only "fine tuning" control you've got is the exposure compensation dial, which works in 1/3 stop increments; I'm sure the difference, if any, is smaller than that!

I tried mounting both lenses, focusing them at infinity, setting them at maximum aperture, and pointing them at a nearby, evenly-illuminated surface, then making an exposure at the meter-determined value. I thought that since the shutter speeds are set steplessly, I might be able to determine a difference in effective aperture by comparing the actual shutter speeds the camera chose (accessible from the EXIF header.) But, no soap -- the meter chose exactly the same value for both lenses.


Next attempt: I know that ISO film speed ratings are based on the least exposure that produces a certain level of shadow detail, so I decided to try a similar methodology on this. I opened my two test frames (exposed at exactly the same shutter speed, remember) and examined them using Photoshop's Histogram palette.

For those who didn't know, a histogram is a graph that shows how many pixels have each brightness level, on a scale of 0 to 255. So, if I found the lowest brightness level that had any pixels, that would be the digital equivalent of ISO's "minimum effective exposure" and should show which lens let in more light.

Sure enough, the Canon 50/1.4 reached its extinction point at a brightness value of 34; there were 4 pixels having this value. Everything below that had 0 pixels. Meanwhile, the Nokton reached its extinction point at level 41, where there were 5 pixels.

However, that's only a difference of 2.7%, which is probably inside the limits of experimental error. ("Never trust a graph without error bars," as one of my professors used to say.) If you decide to accept the figure as accurate anyway... a lens that transmits 2.7% less light than f/1.4 has a nominal aperture of about f/1.42, so by that measure the 50/1.5 Nokon is actually performing relatively better than its aperture rating indicates, which actually wouldn't be surprising in view of its having multi-layer coating (which increases light transmission.)

So, all I can say from this crude measurement is that while the effective transmission of the Canon appears to be higher than that of the Nokton, the difference isn't much, and probably isn't photographically significant.

[I'm not sure about that f/1.6 figure for the Nokton anyway. I've read it in something Sean Reid wrote, but he didn't explain how he came up with it. I'm dubious of his methodology for some of his other exposure comparisons -- too many of them seem to be based on comparing a camera's meter readout vs. the readout he gets from his Sekonic meter. "X is 1/3 stop off because it differs that much from this other meter I happen to have, pointed at whatever light source is handy" isn't my idea of a repeatable test procedure!]