[stevebrot]

I measured a central area (376x432 - just for the simple math) of the raw files using rawshack and compared the Mean values (for Green_2, others are proportional) and also checked the visual results. In all four cases, f/2.8 came out visually darker than the other two (f/4 and f/5.6), which were very consistent (0.1%, 0.1%, 2.5% and 1.6% difference) to each other in each series. The f/2.8 exposures had 29% less (or -0.5EV) at infinity and 24% (-0.4EV) less at MFD, relative to the f/5.6 exposure - very consistent between life view and viewfinder.

Question...Same set EV* for each exposure? (E.g. f/2.8@1/500s, f/4.0@1/250s, f/5.6@1/125s)


Steve

* EV in the original sense of being the actual exposure (x aperture for y time)

[Chosa]

I shot 4 different short tests series against a reasonably evenly illuminated white door, all in center-weighted average measurement mode, at f/2.8, f/4.0, f/5.6 respectively:

1. infinity, live view
2. infinity, viewfinder
3. MFD, live view
4. MFD, viewfinder

I measured a central area (376x432 - just for the simple math) of the raw files using rawshack and compared the Mean values (for Green_2, others are proportional) and also checked the visual results. In all four cases, f/2.8 came out visually darker than the other two (f/4 and f/5.6), which were very consistent (0.1%, 0.1%, 2.5% and 1.6% difference) to each other in each series. The f/2.8 exposures had 29% less (or -0.5EV) at infinity and 24% (-0.4EV) less at MFD, relative to the f/5.6 exposure - very consistent between life view and viewfinder.

For a comparison, I did the very same test with my D-FA 100 WR. At infinity, both the f/2.8 and f/4 exposures came out darker - by 10% (-0.16EV) and ~7% for both LV and viewfinder than the f/5.6. At MFD, only f/2.8 was darker, by 29% (ringing a bell?), and f/4 was actually brighter (~4.5%) than the respective f/5.6 exposures. Overall, the MFD exposures were significantly brighter than the infinity ones.

Somewhat surprised, I double checked the results visually using geeqie's 4-way view, zoomed in to the center at 1:1 to rule out vignetting effects. Visually, the appearance matched what I have calculated. If I find the time, I may put together some scripting to automate evaluation of the results and generate graphs for some of my lenses and maybe compare K-1 and KP.


In conclusion, there doesn't seem to be anythng out of the ordinary with my Irix 150mm.

Watching inside at the aperture diaphragm blades of your irix lens have you noticed any difference between f2.8 and f3.2?
As I said, on my copy, blades activate at F3.2, but so little that they are not visibile, and in manual mode the exposure is identical with f2. 8 or f3.2
My 100wr macro doesn't act like this f3.2 is darker than f2.8 and I can see the aperture blades.

[JensE]

Watching inside at the aperture diaphragm blades of your irix lens have you noticed any difference between f2.8 and f3.2?

They are engaging and get visible - as backed by the measurements. The initial step does look smaller than the following ones, though.

Question...Same set EV* for each exposure? (E.g. f/2.8@1/500s, f/4.0@1/250s, f/5.6@1/125s)

No, this was just a quick test using aperture priority - relevant when you want to know how the camera exposes with a particular lens at a given aperture. Checking the Light Value in the EXIFs, it does vary. I'll redo with constant EV tonight, when it is dark enough for controlled constant light.

[JensE]

I redid the above test using more F-Stops, but only for minimum focusing distance and viewfinder, keeping exposure constant but closing up aperture in 1/3EV steps and increasing exposure time in 1/3EV each step. Results are displayed in a suitable log-log plot below.
I did a single test run only, but it gives you an idea about consistency of exposure once closed far enough, about ±0.1EV, whereas f/2.8 is about 1/3EV below average and f/3.2 is almost in the error margin already.
(Just to make this clear - the vertical scale has an arbitrary starting point, '0' is not special. I could have put it from 5.0 to 6.0 - it's just meant to show the relative exposures in EV scale - i.e.log2-based.)

Attached Images

 

---

Last edited by JensE; 06-10-2020 at 03:52 PM.

[swanlefitte]

What is the light source and shutter speed in case AC cycles can be an impact. At shutter of 1/120 and 60 cycles it only gets half a cycle.

[JensE]

Should be OK - high quality electronic ballasts, T-5 tubes, shutter 1/6s and slower. With electronic shutter, I can see the >20kHz flicker, but no slower changing component.

[JensE]

Maybe I should add that the above 1/3 EV is nothing to worry about for me at all. For critical JPG use, I would need to check the histogram anyway and for everything else, a +1/3EV more of "exposure" (raw value multiplier) doesn't matter at all, so I can use the Irix 150mm with the same exposure settings at f/2.8 as for all other settings.

[Chosa]

I've send it back and take another copy of the lens...
which behave similarly to the first one (the serial n. is different)

It could be the case that I'm particularly unlucky, but I think that this lens have some generic error in the aperture calibration.

[JensE]

I had totally forgotten about what I posted in this thread ! I meanwhile conducted a 4-way comparison of macro lenses at 1:1 (posted on Makro-Vergleich LAOWA 100, DFA 100, IRIX 150, Elicar 90 ? Pentaxians in German) and found that I indeed needed to stop down the IRIX about one stop further than the others, in order to achieve equal depth of field. But it did produce one more shallow step of depth-of-field when going from f/4.0 to f/2.8, shallower than the LAOWA 100mm macro does at f/2.8, so the IRIX's f/2.8 is still considerably brighter than the LAOWA's and the DFA100WR's f/2.8!

[MJKoski]

If you have the guts to do it, open up the lens aperture lever mechanism and see that the ring to which the aperture lever is attached to rotates freely. My 15mm Dragonfly lost the tension of its aperture springs due to huge friction which sometimes jammed up the aperture and it didn't stop down far enough. I had to cut the springs short two times (to gain some tension) and then I replaced the springs with one taken from an old broken K-mount prime. The problem is this:



Click to open a short video. In the beginning of the video the lever is jammed up half way. Then, eventually, it gets stuck again. This causes inconsistent exposures (overexposure) randomly. It also causes K-1 to lose track of the aperture on top lcd.

[stevebrot]

I had totally forgotten about what I posted in this thread ! I meanwhile conducted a 4-way comparison of macro lenses at 1:1 (posted on Makro-Vergleich LAOWA 100, DFA 100, IRIX 150, Elicar 90 ? Pentaxians in German) and found that I indeed needed to stop down the IRIX about one stop further than the others, in order to achieve equal depth of field. But it did produce one more shallow step of depth-of-field when going from f/4.0 to f/2.8, shallower than the LAOWA 100mm macro does at f/2.8, so the IRIX's f/2.8 is still considerably brighter than the LAOWA's and the DFA100WR's f/2.8!

I might take that as evidence that the taking focal length was other than 150mm due to focus breathing at close distance (unavoidable with internal focus lenses). A side-effect is a change in magnification and effective relative aperture from what might be expected. Either might be enough to account for the difference in blur.


Steve

[JensE]

I might take that as evidence that the taking focal length was other than 150mm due to focus breathing at close distance (unavoidable with internal focus lenses). A side-effect is a change in magnification and effective relative aperture from what might be expected. Either might be enough to account for the difference in blur.

I checked the magnification: The scale was correct on all lenses, field of view in the focal plane was the same across all lenses, which I had measured to actually be 36mm on FF before. So the change can indeed be attributed to the change in aperture due to the change in focal length. The Irix has ~93mm FL at 1:1, if I remember correctly (data from Irix at Photokina), and the 'brighter' aperture resulting from it indeed matches the observed shallower depth of field, compared to e.g. the unit-focusing Elicar, which retains both its 90mm focal length and aperture size. The absolute value of the focal length has a negligible impact on the depth of field around 1:1, for a given acceptable blur radius the geometric depth of field (not considering diffraction) is basically just a function of magnification and aperture.

[JensE]

Just very briefly - here are the effective aperture curves for the macro lenses, which I compared for various magnifications, based on measuring the transmitted light in the center (1% - 1/10th in each direction, green channels averaged). Just one measurement series for each lens, so a bit of variation in the mechanical aperture coupling can be expected. Black is the theoretical curve for an ideal lens. I moved the curves vertically (by multiplication), so that they scale from around the same point. The infinity values (magnification = 0) behaved a little erratic, but I was mainly interested in the scaling with respect to magnification. You can see how the Irix and DFA50 appear close to twice as bright as a unit focusing lens would.