Bolex H16 RX Light metering

[carllooper]

This was going to be a question about the effective exposure time of a Bolex H-16 RX but I ended up with the answer instead.

According to the following paper:
http://www2.webster.edu/acadaffairs/asp ... x.H-16.pdf

the Bolex H16 RX has a 135 degree shutter, and 20% to 25% of the incoming light is lost to the reflex viewfinder.

Now normally the exposure time can be calculated as the reciprical of the frame rate x shutter ratio x remaining light ratio.

1/24 x 135/360 x 80/100 = 0.0125 secs = 1/80 secs [24fps, 135 degree shutter, 20% light loss]
1/24 x 135/360 x 75/100 = 0.0117 secs = 1/85 secs [24fps, 135 degree shutter, 25% light loss]

But the paper says that Bolex have determined the effective exposure time for RX lenses on RX cameras, at 24fps, is actually 1/65 secs !

This is because the aperture markings on RX lenses are not actually the correct aperture markings. The markings on RX lenses compensate for the light lost to the viewfinder. In other words, when calculating effective exposure time one needs to re-compensate for the RX lens compensation by removing from the otherwise correct equation, the light loss ratio and assume 100% light throughput.

When setting the light meter, the equation for effective exposure time on RX lenses then becomes:

1/24 x 135/360 = 0.015625 secs = 1/64 secs [24fps, 135 degree shutter, 0% light loss]

which is more or less in agreement with what Bolex are saying.

Carl

[bolextech]

As it has been discussed in this forum in the past, whoever wrote those "instructions" has got it totally wrong.
Your original calculation that gives an exposure time of 1/80 for 24fps is the right one.
Just check out any Bolex user's manual for a reliable table of adapted exposure times.
The RX designation on lenses has absolutely nothing to do with the diaphragm opening.
It just means that the lens is optically compensated for the beamsplitter prism in terms of focus.

Cheers,
Jean-Louis

[joelpierre]

http://cjoint.com/13jv/CABuWGApger_bole ... e_time.pdf

[carllooper]

As it has been discussed in this forum in the past, whoever wrote those "instructions" has got it totally wrong.
Your original calculation that gives an exposure time of 1/80 for 24fps is the right one.
Just check out any Bolex user's manual for a reliable table of adapted exposure times.
The RX designation on lenses has absolutely nothing to do with the diaphragm opening.
It just means that the lens is optically compensated for the beamsplitter prism in terms of focus.

Cheers,
Jean-Louis

I don't know about that.

The real exposure time is 1/65 second since the lens can't alter the real exposure time. But if there is light loss (and there is) then you would have to set the exposure time on the light meter to 1/80 sec, to accomodate for that light lost to the viewfinder. The light meter assumes a camera without light loss to the viewfinder.

But if the lens does in fact correct for the light loss then you would have to set the light meter back to 1/65 sec.

How does the lens correct for the light lost to the viewfinder? Does it pump 20% more light onto the film compared to other lenses?

If it corrects for the light loss it can only do so by altering the f/number on the barrel of the lens with respect to the real aperture. For example, where it says on the barrel that the aperture is f5.6, it would have to, in real life, be a fraction of a stop closer to f/4, ie. to let more light in with respect to the f/number. No doubt the idea of the engineers being to make calculation of the light meter setting easier (in the days before handheld calculators). In other words one wouldn't have to include the light loss factor.

So one should set the light meter to 1/65th second (rather than 1/80th second), to take into account the correction the lens has already made. In other words one would have to remove from one's equation an account of the light loss, since the lens has already accomodated for that loss.

To put it another way, only if the lens was not already compensating for the light loss (using 'fake' f/numbers), would you set the light meter for 1/80th sec.

Surely?

What I don't know is if an RX lens does in fact already compensate for the light loss (using 'fake' f numbers). If the correct light meter setting really was 1/80th second, then the lens would have to be regarded as not already correcting for the light loss since the calculated 1/80th sec is for accomodating the light loss. But if the lens does do a correction (using 'fake' f numbers) then the correct light meter setting would be 1/65 sec (or 1/64 sec according to my calculation).

Carl

Explanation of the equation.

At 24 fps, a shutterless system would have an exposure time of 1/24th sec per frame (ignoring pull through). The shutter, however, provides only 135 degrees out of 360 degrees, meaning it cuts exposure time down to 135/360 (or 37.5%) of an otherwise shutterless exposure time.

So the real exposure time per frame is:

1/24 x 135/360 = 0.015625 secs = 1/64th sec

That's how long the film is being exposed to light. If the camera loses no light to the viewfinder then using a light meter that assumes the same (as light meters do) then you would set the light meter to 1/64th sec (or perhaps 1/65th sec to account for light lost by the lens glass).

Light meter reports an aperture to use. And you set the aperture.

But if the camera loses light to the viewfinder (as it does) the light meter doesn't know about this. It is reproting an aperture to use based on no light loss. So you would (normally) have to accomodate for this. If 20% (20/100) light is lost to the viewfinder, then only 80% (80/100) of the light that would otherwise reach the film is reaching the film. So you tell your light meter that the exposure time is actually shorter, by that amount:

1/24 x 135/360 x 80/100 = 1/80th sec

The light meter reports an aperture to use, as if you were using a shutter of 1/80th sec without light loss, which is the same as 1/64th sec with light loss.

BUT - if the lens is already performing this correction then you don't do this. You end up back with:

1/24 x 135/360 = 0.015625 secs = 1/64th sec

Or perhaps 1/65th second to account for the minor light loss by the lens glass.

Carl

[bolextech]

Carl, you're not listening. Reread carefully what I said. Look at the table posted by joel pierre. It says it all there. Real and adapted.

Cheers,
Jean-Louis

[carllooper]

Hi Jean Louis,

I'm listening. But I don't trust the tables. What does "real" and "adapted" mean? I take it to mean the real exposure time (in agreement with what I calculated) is 1/65th second. But one must adapt that to 1/80th second, to accomodate light loss, also in agreement with my calculation. In other words I don't need the table. My equations say exactly the same thing. As they should since the equations are no doubt the basis for the table.

The issue I have is whether RX lenses (as per the cited paper) correct for the light loss or not? If not then the correct exposure time (for the light meter) is 1/80th sec, as per the table and my calculations. But if the lenses do (perversely) correct for light loss, then the correct exposure time for the light meter is back to 1/64th sec (or 1/65th sec)

I also understand (after re-reading your post) what you mean by the RX designation on a lens being in terms of focus only, ie. that an RX lens doesn't use 'fake' f numbers.

But do you know this for sure? I mean, if you do, I'll take your word for it.

The reason I ask is that there is competing information on the web. Some commentators refer to this discussion as "not that furfy again" but don't really explain themselves. So how does one trust them? Others just state the opposite as a matter of fact, and again how does one trust them (where do they get info?)

For example, the following is from the University of Texas, https://wikis.utexas.edu/display/comm/B ... mera+guide

The optical system of the Bolex H-16 reflex cameras utilizes a beam splitter permitting through the lens viewing at all times free from flicker. The reflex viewfinder enables accurate focusing and framing, and allows you to estimate the depth of field. The reflex prism deflects 20 - 25% of the light passing through the lens into the viewfinder system. Only 75 – 80% of the light reaches the film plane. The actual quality of the light reaching the film is reduced by about 1/2 to 1/3 of an f-stop. To compensate for the light reduction, Bolex has determined that the effective shutter speed for the H-16 camera is 1/80 of a second rather than the standard 1/65 of a second.

Another compensation for the light loss is the RX lens designed by Bolex with Kern/Switar.The RX lens is calibrated to pass 1/2 to 1/3 more light than the aperture markings on the barrel indicates. When using RX lenses with the H-16 camera, the effective shutter speed is back to 1/65 of a second.



Is this just a myth that has taken hold somewhere and been perpetuated (as some commentators suggest) - ie. that RX lenses admit more light than their aperture markings indicate? Or is it true? And if true where is the proof? Or reciprically, if not true, where is the counter proof? In the absence of either I guess the only answer is to work out some test for the lenses and find out.

Rather fortuitiously (due to ignorance on my part when purchasing lenses) I ended up with two Kern switar 10mm lenses, one of which is RX and the other which is not. So I guess the proof of the pudding is to shine some light through both onto a light meter and see if there is 1/3 to 1/2 stop difference when they are both on the same f/stop.

TEST RESULTS

I set the aperture (diaphragm) of both lenses to f/1.6
With each lens pointing at the same light source I measured the light coming through both.

The answer?

The RX lens admitted 1/2 stop more light. No question.

[bolextech]

You can trust the information provided by Bolex themselves.

Glad you pointed out that Texas information. Another example of people getting it wrong.
The second paragraph starting with "Another..." is wrong and should be disregarded.

You will not see anything in the Bolex literature mentioning anything about the f/stops being "compensated" for on RX lenses.

Read this:
http://www.apecity.com/manuals/pdf/bole ... ameras.pdf

and this:
http://www.city-net.com/~fodder/bolex/truth.html

Finally, physical testing of identical focal length (in this case: 25mm) RX and non-RX lenses on a calibrated light source with light transmission measuring sensor shows no difference at all.

Cheers,
Jean-Louis

[carllooper]

Neither of those papers say one one way or another whether the RX lens admits more light, or not, with respect to the aperture settings. As they say in science, the absence of proof is not proof.

The physical test I just did, however, showed a definite 1/2 stop difference between the RX and non RX lens. There was absolutely no doubt. The RX lens admitted more light than the non RX for the same barrel markings and same light source.

What setup are you using? I'm not sure why we're getting different results. I placed each lens flat against my computer screen on the same white display. I opened the apertures on both to their widest, which is marked f/1.6, on both. Using the reflected light cap on my light meter, I put the meter against the lens. Nice snug fit on both. And took readings.

Interestingly if I closed down the aperture on both the difference in readings started to drop. At a setting of f/4 on both the difference in light had dropped to 1/3 stop (but a significant difference nevertheless). I'll try some more rigourous testing to see what's happening there. I'm assuming that as the aperture is closed, the incursion of black is diluting the difference. A more rigourous test would be to measure the light at the imaging plane on each, ie. making sure to avoid measurement of the out of field that might pollute the results.

Of course, in this day and age, with film scanning and digital grading it hardly matters a difference of 1/2 stop. More out of technical curiosity than anything else.

Carl

[carllooper]

http://cjoint.com/13jv/CABuWGApger_bole ... e_time.pdf

I see you've posted this table in cinematography.com against the same debate. But from what I can tell this is from a manual that was printed almost a decade earlier than the introduction of RX lenses.

So it doesn't really mean anything. The issue is not the camera but whether RX lenses have T stops printed on the barrel, or if they have F stops.

If they are T stops (as some sources suggest) then the light meter needs to be calibrated at 1/65 sec.
If they are F stops then the light meter needs to be calibrated at 1/80 sec (as per original camera manual).

This follows from the mathematics for a physical system specified as 24 fps, 135/360 shutter, 80/100 light transmission, F stop and T stop marked lens) rather than blind faith in a camera table.

From the physical test I did the markings on the RX lens would appear to be in T stops. And if so then the light meter needs to be set for 1/65th sec for the specified system, regardless of what any manual or any other source says.

But I'll hold off arguing definitively that RX lenses are marked in T stops until I do more tests. The physical test I did could very well be some coincidence (although I can not for the life of me explain what would cause such a coincidental difference in transmission other than that the RX really is marked in T stops, and the other in F stops).

Carl

[carllooper]

So using a number of G clamps, a dismantled Bolex (from my collection of such) to ensure the exact same positioning of the lenses, and more accurate positioning of the light meter, I did another test in which I used a bare light sensor (without any head) placed in the lens image plane, with the entire image frame projecting directly onto the bare sensor, and I'm getting the same result, a 1/2 stop difference between the RX lens and the non RX lens.

So I'm pretty sure now the so called myth that RX lenses are marked with T stops is not actually a myth at all, and that RX lenses - or at least the one I'm testing - is indeed marked with T Stops. In other words you would indeed calibrate the light meter at 1/65 sec, instead of 1/80 sec, (for 24 fps, 135/360 shutter and Bolex reflex camera).

Or to be more open about this I'm not sure how any other conclusion can be reached. What would otherwise account for the consistently half stop difference? But given the controversy I recommend anyone interested in this not take my word for it and try testing any RX vs non RX lenses they might have, for themselves. Maybe it was only some RX lenses that were marked in T stops. And I happen to have one. In terms of the historical record, accessible via the web, it remains a mystery. There isn't any reasonably reliable records I've found that might otherwise settle it.

The only other test I can do (and will do) is shoot some film using both lenses, set at the same aperture, for the same scene, and see if there is any difference in exposure. That'll be the clincher, if only for the particular RX 10mm I'm using.

Carl

ps. I've sent a message to Bolex in Switzerland to see what they might be able to unearth.

[carllooper]

http://cjoint.com/13jv/CABuWGApger_bole ... e_time.pdf

I take back what I said about this manual earlier. The camera depicted in the manual (from which the table is lifted) is at least a REX-3 (the actual model I've got) which is around about 1963/1964 from what I can tell - so it's more like 4 years earlier than the 10mm RX I'm looking at, rather than the 10 years I was assuming earlier.

And interestingly, in a photograph of one of the lenses in the manual, you can just make out an RX designation on the lens if I'm not mistaken.

So there is definitely a conundrum here.

Carl

[carllooper]

I did another test today using a different method.

This time I attached the light meter (a digital sekonic), to the eyepiece of the Bolex, firmly fixed in place with clamps. The camera was put on a good sturdy tripod and set pointing at a diffuse light source. No movement.

And this time I tested each lens, for all the stops marked on each lens: from f/22 to f/1.6

The result was exactly the same for all lens stops.

The light meter reported the RX as admitting exactly 0.4 stops more light than the non-RX (for the same lens markings). I used the EV readout on the light meter to obtain the fractional reading.

I've also the searched the web high and low for any historical traces regarding this. Old magazine articles etc. Advertising, etc. Zero. Nothing. All I've got is a 0.4 stop difference and no explanation - other than the quite reasonable (if unsubstantiated) one that this particular RX lens was indeed marked in T stops that took into account the light loss to the viewfinder.

But as evidenced by the otherwise beautiful manual, depicting an RX lens between it's covers, and a table that describes lenses in terms that can't be anything other than lenses marked in f/stops, the RX lenses of that day must be in f/stops.

The only thing I can think of is that during the late sixties (when the preset I've got was made) they changed their policy on RX lenses and either forgot to tell anyone, or the historical record on such is hidden somewhere, yet to be unearthed. There's a lot going on in the 60s. TTL metering is becoming fashionable. There's a whole emphasis on making cameras user-friendly, ie. for the non-technical user. Indeed a lens marked in stops that takes into account light loss to the viewfinder can be regarded as a form of "user-friendliness" even if it's causing us questions today. There's a certain high modernism occuring in the sixties- out with the old and in with the new - rather than necessarily any concerns with backward compatibility, as we might have today.

Anyway. That's the end of my studio tests. I'll post again when I've shot some film. Despite the convincing studio tests I won't be entirely convinced until I see the difference on actual film.

Carl

ps. I read somewhere the Bolex I've got (a REX 3) actually has a 133 degree shutter, rather than a 135 degree shutter. When I plug 133 into the equations (rather than 135) I get 1/65th sec as the real exposure time, rather than 1/64th sec, ie. more in agreement with the table printed in the Bolex manual.

pps. the other thing I've noticed, just eyeballing through the glass at the aperture on both lenses, is that the aperture on the RX even looks bigger than the one on the non-RX, ie. for the same setting on both. It's easier to see when the apertures are set at f/2 than at f/22. Could be just some optical effect of course, or some subjective prejudice influencing my eye.

[cameratech]

Well I have to admire your "test it myself" tenacity even if it ties you up in knots!

But you really should listen to Jean-Louis, he knows more about Bolexes than just about anyone on the planet..

To be perfectly clear, RX lenses do not compensate for the prism light-loss. The aperture marks are f stops, mathematically derived from the geometric relationship of the lens focal length to entrance pupil diameter. Other lenses which may be marked in T stops are adjusting for light loss through the lens itself, usually fractions of a stop slower. The RX designation just means a lens is optically corrected to work with the prism, dealing with the introduced aberrations - it's about the picture quality, not light transmission.

If your tests are finding discrepancies between RX and non RX lenses there are several possibilities. Your testing method may be flawed. One or both iris rings may be mis-positioned. One lens may be newer and have better coatings, meaning less light loss. Or the other lens may have coating damage or yellowing or some other fault that is causing more light loss. Most likely a combination of the above.

It's worth remembering that a case study of 2 individuals isn't a very reliable basis to generalise about a whole population.

[carllooper]

you really should listen to Jean-Louis, he knows more about Bolexes than just about anyone on the planet..

Of that I have no doubt. This isn't about not listening. I'm listening. Indeed I'm listening to everything, which may be my problem of course.

To be perfectly clear, RX lenses do not compensate for the prism light-loss. The aperture marks are f stops, mathematically derived from the geometric relationship of the lens focal length to entrance pupil diameter.

Its not as if I don't understand what Jean-Louis is saying. My ears are working. That's not the problem.

Other lenses which may be marked in T stops are adjusting for light loss through the lens itself, usually fractions of a stop slower.

Yes, I know. It does sound highly unusual that a lens would be marked to adjust for light loss through anything other than itself. On the other hand the RX lenses were specifically made for the Bolex.

The RX designation just means a lens is optically corrected to work with the prism, dealing with the introduced aberrations - it's about the picture quality, not light transmission.

Yes. That is what Jean-Louis said. I get that.

If your tests are finding discrepancies between RX and non RX lenses there are several possibilities. Your testing method may be flawed. One or both iris rings may be mis-positioned. One lens may be newer and have better coatings, meaning less light loss. Or the other lens may have coating damage or yellowing or some other fault that is causing more light loss. Most likely a combination of the above.

Thanks Dom, I'm quite happy to accept this as a possibility. Could well be correct. But I'm not convinced the alternative isn't so either. Jean-Louis provided me the reasons why he has adopted his position, and they are very good reasons. In other words I don't have take Jean-Loius word for it, I can listen to the extremely good reasons he has supplied:

1. A Bolex manual is printed with a table consistent with lenses marked in f/stops
2. A Bolex tech document on the RX doesn't mention any accommodation for light loss
3. A physical test he did between an RX and non-RX 25mm lens did not produce any difference in light.

I'm very persuaded by this but obviously not entirely so.

It's worth remembering that a case study of 2 individuals isn't a very reliable basis to generalise about a whole population.

Yes, I totally agree. Much of what I've said in the course of this thread is towards that exact same conclusion.

cheers
Carl

[carllooper]

Other lenses which may be marked in T stops are adjusting for light loss through the lens itself, usually fractions of a stop slower.

Thinking about this more, if (and I say if), if the RX lens I'm testing was indeed marked in T stops it would make no sense to mark it in terms of the lens alone as this would defeat the purpose of T stops. The purpose is to avoid having to open the aperture a tad (by different amounts for each lens) relative to the light-to-f/stop measurement/calculation performed by the light meter. While this is more useful for zoom lenses rather than primes, the Bolex reflex is a slightly different beast. The light lost to the viewfinder is part of the complete transmission system. Since an RX is specifically designed for the Bolex, and less than ideal on any other system, it makes sense to mark it in terms of the Bolex specific transmission system.

In other words, if (and I say if), if a particular batch of prime RX lenses were to be manufactured with T stops they should take into account the light lost to the viewfinder. Otherwise there would be no purpose since a prime lens on it's own would have insignificant differences in terms of transmission.

One point you've raised is good. If the transmission difference I'm reading really is a result of deterioration and/or misalignment, etc. a film test isn't going to alter how we might otherwise interpret the results. There will still be a question. From a practical point of view it doesn't really matter of course. A film test will set up the basis for what corrections to use regardless of cause.

Carl