Lomo 3.7x finite vs 120G

[chris_ma]

I would worry about not using the same lens for the tests. I've seen fairly significant differences in exposure between lenses of same aperture. T-stops?

agreed, the F-stop T-stop difference could introduce some errors when trying to calculate the physical aperture - but since I don't see a way to do it without a reference lens there's not much we can do about it, and on most modern coated lenses the light loss is fairly consistent.

Anyway, using same lens would fix that, but then I'd need to rely heavily on the curve. I'm not sure how to linearize the demosaicing/raw conversion.

I don't see how you could do it with the same lens - I thought we were talking about stacked lenses of unknown aperture here and the idea is to use another lens with known aperture for exposure comparison and calculate the F-stop (or more precise T-stop) of the unknown lens based on the exposure difference?

an easy way to linearize the RAW data values seems to be maketiff (I only used it briefly because my workflow usually involves Lightroom):
https://www.colorperfect.com/MakeTiff/

[ray_parkhurst]

I measured the exit pupil of the combination at 10mm. The extension from rear of lens to sensor is 110mm, so I believe the EA is f11.

Now the question is if I measured the exit pupil correctly. I held the lens at arm length, and looked through it at my monitor while holding a ruler up to the back of the lens, measuring the diameter of the bright circle. Does that give a correct number?

That's close, but you also need to know the distance between exit pupil and sensor.

You can accurately determine both size and distance by looking into the rear of the lens, using a camera equipped with a macro lens. Focus and photograph the exit pupil, then remove the lens under test, swap in a ruler placed so that it's perfectly focused too, and take a second photo of that.

The ruler now gives you both the location and the size of the exit pupil, so you can calculate NAatSensor = RadiusOfExitPupil / DistanceFromPupilToSensor in the target system.

Effective F-number is then just Feff = 1/(2*NAatSensor) as usual.

--Rik

So I think the only error is a small bit a parallax due to perspective of my eye vs the exit pupil and ruler, correct? Wouldn't I need to use a telecentric lens to measure the exit pupil exactly? I actually "saw" a little more than 10mm, and rounded down to compensate a little for the parallax. I did place the ruler right on rear of lens, and then measured extension from there to the sensor, so I think that part is correct.

Or will there be significant error due to the focusing of the aperture?

[ray_parkhurst]

I don't see how you could do it with the same lens - I thought we were talking about stacked lenses of unknown aperture here and the idea is to use another lens with known aperture for exposure comparison and calculate the F-stop (or more precise T-stop) of the unknown lens based on the exposure difference?

It would be easy to adjust the objective to have same mag as the stacked combo. The objective is known to have f6.3, so it would just be a delta from there.

[rjlittlefield]

I measured the exit pupil of the combination at 10mm. The extension from rear of lens to sensor is 110mm, so I believe the EA is f11.

Now the question is if I measured the exit pupil correctly. I held the lens at arm length, and looked through it at my monitor while holding a ruler up to the back of the lens, measuring the diameter of the bright circle. Does that give a correct number?

That's close, but you also need to know the distance between exit pupil and sensor.

You can accurately determine both size and distance by looking into the rear of the lens, using a camera equipped with a macro lens. Focus and photograph the exit pupil, then remove the lens under test, swap in a ruler placed so that it's perfectly focused too, and take a second photo of that.

The ruler now gives you both the location and the size of the exit pupil, so you can calculate NAatSensor = RadiusOfExitPupil / DistanceFromPupilToSensor in the target system.

Effective F-number is then just Feff = 1/(2*NAatSensor) as usual.

--Rik

So I think the only error is a small bit a parallax due to perspective of my eye vs the exit pupil and ruler, correct? Wouldn't I need to use a telecentric lens to measure the exit pupil exactly? I actually "saw" a little more than 10mm, and rounded down to compensate a little for the parallax. I did place the ruler right on rear of lens, and then measured extension from there to the sensor, so I think that part is correct.

Or will there be significant error due to the focusing of the aperture?

No, you do not need a telecentric lens to take the measurement. The procedure I describe is just using the macro lens to make an accurate measurement of the location and size of a virtual image, to wit, the virtual image of the physical aperture as it is seen through the rear of the lens. By definition, that virtual image is the exit pupil. It has a particular location (typically somewhere inside the lens, but maybe deep inside or even somewhere outside it).

In the arm's length method, if you're vulnerable to parallax error than you're also guaranteed that the pupil is not located at the rear of the lens, so you'll have a distance-from-sensor error also.

--Rik

[ray_parkhurst]

You can accurately determine both size and distance by looking into the rear of the lens, using a camera equipped with a macro lens. Focus and photograph the exit pupil, then remove the lens under test, swap in a ruler placed so that it's perfectly focused too, and take a second photo of that.

The ruler now gives you both the location and the size of the exit pupil, so you can calculate NAatSensor = RadiusOfExitPupil / DistanceFromPupilToSensor in the target system.

Effective F-number is then just Feff = 1/(2*NAatSensor) as usual.



In the arm's length method, if you're vulnerable to parallax error than you're also guaranteed that the pupil is not located at the rear of the lens, so you'll have a distance-from-sensor error also.

--Rik

OK, I used the macro lens method, and although the source numbers are very different, the result is similar.

Exit pupil diameter is 19mm vs the 10mm for arms-length method

Distance from exit pupil to sensor is 202mm vs 110mm for arms-length method

NA is (19/2)/202=.047
EA = 1/(2*NA)=10.6

So I would guess that if you could accurately measure the exit pupil without parallax at any point in the air space between the lens and sensor, you would come up with a similar answer.

edited to add: what I mean is if you could accurately measure the apparent size of the exit pupil

[rjlittlefield]

So I would guess that if you could accurately measure the exit pupil without parallax at any point in the air space between the lens and sensor, you would come up with a similar answer.

edited to add: what I mean is if you could accurately measure the apparent size of the exit pupil

What you're really trying to figure out is the angle subtended by the exit pupil, as seen from a point on the sensor.

I'm not quite sure what you've said about the measurements, but if I've guessed right then it's correct in the special case that your eye happens to be located at the same distance from the lens that the sensor will be.

If your eye is not located at that critical distance, then you'll get different values, and the farther you are from the critical distance, the more different the values will be.

--Rik

[ray_parkhurst]

So I would guess that if you could accurately measure the exit pupil without parallax at any point in the air space between the lens and sensor, you would come up with a similar answer.

edited to add: what I mean is if you could accurately measure the apparent size of the exit pupil

What you're really trying to figure out is the angle subtended by the exit pupil, as seen from a point on the sensor.

I'm not quite sure what you've said about the measurements, but if I've guessed right then it's correct in the special case that your eye happens to be located at the same distance from the lens that the sensor will be.

If your eye is not located at that critical distance, then you'll get different values, and the farther you are from the critical distance, the more different the values will be.

--Rik

That makes sense. So if rather than holding the lens at arms-length, I held it such that my eye were at the sensor plane, then there would be no parallax problem skewing the measurement and the apparent aperture size should be correct. That would jibe with my arms-length result, which indicated a slightly smaller aperture reading than I would have made if my eye were closer.

[RobertOToole]

I shot the Lomo 3.7x in both finite and infinite modes to compare performance. As we all know, if indeed the infinite tube lens complements the objective well, the performance "should" be better based on the FL multiplication making a slightly wider effective aperture. In this case, the improvement is 1-3.7/4.7 = 21%. However, with the 120G, I see both improved sharpness and contrast. Makes me wonder if the Lomo was designed for infinite use? Anyway, here are the comparison shots. You won't see much difference in the overall, but there are subtle but noticeable differences in the center and corner crops, with the 120mm Gretag tube lens showing both improved contrast and sharpness.

Glad to see you had some nice results with the Lomo 3,7 with a tube lens Ray.

If anyone is curious, a higher contrast target will show a lot more of a difference in sharpness with easy to see results, I used an SK 5.6/120 Makro-Symmar for my test but the G 120 has a very similar high performance output so the results should be very close in my experience with both lenses.

You can see the Lomo 3,7 x + SK 5.6/120 Makro-Symmar results here on the forum, the thread that is referred to a couple of times in this post, FYI:

https://www.photomacrography.net/forum/ ... hp?t=38701

and the test on my site:

https://www.closeuphotography.com/lomo-3-7x-and-sr120

[RobertOToole]

You can do it very accurately, though it is time-consuming.

there is another way which would be faster:
since sensors in cameras react pretty linearly to light, if you take a RAW photo and bypass any contrast/gamma curve, your RGB values are linear as well so you can simply find the multiplication factor.

note that most RAW converters will apply a contrast curve to make things prettier, and a gamma curve to make things look good on monitors, so the hardest part is to actually get a linear RAW conversion. I created a linear profile for my S1R for Lightroom and export with ProPhoto profile, then reverse the gamma correction of ProPhoto later which works pretty well.

yet another option would be not to worry and to just use the exposure slider in Lightroom and read the EV difference there - you'd probably get less then 1/3 stop error this way as long as the reference lens has a similar f-stop.

chris

Hi Chris,

There is a lot more to what you touch on in your post. It's a lot worse, more complicated, turns out that "0" with adobe sliders is not even close to a real baseline. Everyone on this forum should be aware of this really.

If you have some free time, browse over to the, always excellent, RawDigger site:

Deriving Hidden Baseline Exposure Compensation Applied by a Raw Converter

https://www.rawdigger.com/howtouse/deri ... mpensation

This is a very informative post, definitely an eye-opener.

Also this is more technical but also good but you will need a lot of time to go over all the info:

Forcing a Raw Converter to Render Tones Accurately

https://www.rawdigger.com/howtouse/over ... s-settings

Be warned that you will spend a lot time reading around on the RawDigger site if it's your firs time.

[chris_ma]

Hi Robert,

thanks for the heads up, and for the excellent links.
I'm aware of the non linearity of RAW converters, specially Adobes, but forgot about the process version and didn't know about the base exposure. those might help to solve some problems I'm having with calibration the camera for my workflow.

I made a camera profile that reacts linearly, I'll compare it to switching to PV 2 (2010).
the base exposure shouldn't really affect results when comparing lenses since it's the same for all. will see that I can do some tests later.

I'm aware that even with these corrections, it's not a totally accurate tool to measure effective aperture, but it should be possible to get it roughly right (say within 1/3 stop) and easier for most people then to measure and calculate it for stacked lens combinations.
chris

[mjkzz]

You can accurately determine both size and distance by looking into the rear of the lens, using a camera equipped with a macro lens. Focus and photograph the exit pupil, then remove the lens under test, swap in a ruler placed so that it's perfectly focused too, and take a second photo of that.

The ruler now gives you both the location and the size of the exit pupil, so you can calculate NAatSensor = RadiusOfExitPupil / DistanceFromPupilToSensor in the target system.

Effective F-number is then just Feff = 1/(2*NAatSensor) as usual.

--Rik

Wow, nice trick.

I think something maybe critical when doing this measurement, just to make sure I understand it right.

First the two optical axes for the macro lens and the lens being tested should coincide (this will also ensure image plane and the exit pupil are parallel to each other). If not, at least try to make them parallel to each other to reduce error.

Second, I guess when doing the second photograph of ruler, do not re-focus the macro lens, keep it as it is (this might be stupid to think, but when doing experiment, it could happen). And the ruler must be placed parallel (perfectly to reduce error) to the same image plane. Then the exit pupil to sensor distance will be ruler to camera mount + flange distance. The size of exit pupil is the ruler reading masked by the image of exit pupil in the first image. This process is essentially saying that the ruler is placed right behind the (infinitely thin) exit pupil and is being photographed. Very, very neat.

Thirdly, I think the distance between the camera with macro lens to the lens being tested does NOT matter, as long as the macro lens is not re-focused. This can be very useful as it eliminates another factor of "being perfect". Focal length and where the macro lens is focused does not matter.

[mjkzz]

just to add, the reading of ruler in second image should be in the center of exit pupil in first photograph. Well, to be sure, if all images are shot centered, this might not be a problem.

[mjkzz]

another thought, it is essentially saying we are taking an image of an infinitely thin PHYSICAL pupil with a ruler behind it . . . so maybe those "perfections" might not even be needed. But with all practical equipment, maybe it is good to have at least some kind of "perfection", maybe shoot it at center of image?

In this case, wow, so nice.

[chris_ma]

did a brief test with stopping down a known good lens and later adjusting the exposure slider in Lightroom to match.
I found the error to stay around 5% within a 3stop range (which could well have been the aperture not being precisely set) no matter what camera profile and process version I used when measuring mid tones.

so I think for rough estimations, taking a picture with a known aperture lens, then taking one with a unknown aperture lens combination and adjusting the exposure slider (or shutter in camera) to match mid tone brightness should give us a pretty accurate result of light transmission. If it's a clean modern lens that should translate pretty well to the effective aperture of the new lens combo.

[ray_parkhurst]

did a brief test with stopping down a known good lens and later adjusting the exposure slider in Lightroom to match.
I found the error to stay around 5% within a 3stop range (which could well have been the aperture not being precisely set) no matter what camera profile and process version I used when measuring mid tones.

so I think for rough estimations, taking a picture with a known aperture lens, then taking one with a unknown aperture lens combination and adjusting the exposure slider (or shutter in camera) to match mid tone brightness should give us a pretty accurate result of light transmission. If it's a clean modern lens that should translate pretty well to the effective aperture of the new lens combo.

Ultimately I think the physical measurement is the best way to go. It was actually pretty easy.