T-stops better: then why f stops?

[David Lyga]

I think that Hollywood, whose exposures have to be correct, have T-stops incorporated into their lenses (taking into account all lens aberrations which compromise full accuracy). T- stops (transmission stops) are calibrated so as to give an actual doubling or having of light transmission with each step. F stops are supposed to also, but they are not perfectly matched to that precise requirement, although theoretically they are. Is there a reason why manufacturers opt for the somewhat less accurate f stop? - David Lyga

 

[MattKing]

Cost, and questions about usability.
A T-stop calibrated lens has to be individually calibrated. That is one of the reasons why the top end movie lenses are so extraordinarily expensive.
And the usability issues arise from the fact that things like depth of field calculations are based on f-stop numbers, not transmission numbers.

 

[Old-N-Feeble]

Wouldn't the accuracy of T-stops be affected by lens flare just as F-stops are?

 

[Nodda Duma]

Aberrations are a mathematical function of f/#, not T-stop.

T-stop for a given lens can vary based on factors external to the optical design itself. f/# itself is set by the design and doesn't change if a filter is added.

Regards,
Jason

 

[David Lyga]

Well, I am learning something. Speak on. - David Lyga

 

[Luckless]

From my understanding the key reason for video production using T-Stops is visual consistency, and the demands simply aren't nearly as tight in photography as videography due to how they're viewed and produced.

The human eye is very good at detecting changes in light. Not great at precisely measuring the change, but it becomes very obvious when there is a change. This is a problem if you're trying to make a movie because you will often not only cut between different camera angles in a scene, but also different takes. Combine different takes with reshoots, and imprecise f/stops become an editing headache.

If you go out to shoot a scene and are shooting at f/8 one day, then you want all the lighting to line up perfectly between your wide angle, your normal, and your telephoto lens shots. You don't want one to be a bit darker and one a bit lighter, as it will be harder to correct in post, and be more difficult to polish off in final editing.

Now to make things even more interesting - You come back to the same location to do reshoots two weeks later, and your camera crew arrives with the lenses handed to them that morning by the rental house... Your video post production team is not going to be happy with the results if the camera crew slapped on a different copy of the same model of lens, and 'surprise!' f/8 on this one isn't really completely the same as f/8 on the previous lens.

A small shift in depth of field? Well, that can be lived with. You're probably not going to notice too much from scene to scene/take to take if there is much motion involved. The human eye just can't notice it as easily. But it will notice if the light level keeps randomly shifting a bit between cuts.

 

[flavio81]

I think that Hollywood, whose exposures have to be correct, have T-stops incorporated into their lenses (taking into account all lens aberrations which compromise full accuracy). T- stops (transmission stops) are calibrated so as to give an actual doubling or having of light transmission with each step. F stops are supposed to also, but they are not perfectly matched to that precise requirement, although theoretically they are. Is there a reason why manufacturers opt for the somewhat less accurate f stop? - David Lyga

f/stops are geometrical. f stop value = focal length in mm / mechanical aperture in mm (or so I've been told)
They do not consider the transmission of a lens. A f2.0 lens with 6 optical groups might transmit less light than a f2.0 lens with 4 optical groups, assuming coatings are equal.

T/stops consider transmission. So a T2.0 lens will always pass through exactly the same amount of light!!

This allows perfect consistency of exposure. I guess this is meaningful when they have to shoot a zoom lens (of many many lens groups) versus a prime lens. I would bet that back in the 60s where lenses were single-coated, this made a big difference in light transmission.

BTW, f/stops on a camera lens are not usually THAT precise...

 

[Leigh B]

No lens element transmits 100% of the light incident on its first surface.

f/stops ignore this, so a 2-element design shows the same transmittance as a 10-element design.
T/stops, being measured rather than calculated, move to show the higher attenuation of the latter.

- Leigh

 

[Jim Jones]

T/stops were appropriate for cine lenses and transparency film. Rarely does a still photographer need such a consistent exposure with a variety of multicoated lenses.

 

[jeztastic]

This video provided some useful info on this subject, and was very entertaining as well...

 

[guangong]

As a sometimes cinematographer the information presented above is perfectly true. I only want to add a footnote:not all cinema lenses are calibrated in T marking, but only those of the highest quality and usually not owned by individuals but by rental houses. Most lenses for the semiprofessional use standard f stops. A twenty, thirty or more thousand dollars for a T marked lens is not unusual and rental fees reflect this cost.

 

[Ko.Fe.]

Why bother and bring cost up, if f-stops are working perfectly for photography?
And actually I have T stops. Av for aperture, Tv for shutter speeds. On my Leica M4-2, I have f-stops on the lens and t-stops on shutter dial

 

[darkroommike]

Why bother and bring cost up, if f-stops are working perfectly for photography?
And actually I have T stops. Av for aperture, Tv for shutter speeds. On my Leica M4-2, I have f-stops on the lens and t-stops on shutter dial

Most pro work in cinematography is still done with hand held metering so no TTL, no Tv or Av settings on cameras. The T-stop was never a thing for RF cameras, nor for still cameras of any type where you can fix it in the darkroom. Exception being for the early cine zooms adapted to still cameras (Angénieux and other). The T stop is a work around adopted in the early days of cine zooms lens. As others have said it is an actual as opposed to a calculated value for the aperture light ratio. Especially with early zooms, with elements having very rudimentary anti flare coatings, flocking, baffles, etc. it was found that some zooms at some settings were only about 85% (average) efficient in getting light through the pipe. Lens coatings, etc. have really improved but for critical professional work time is still money so the less fiddling a cinematographer has to do the better.

 

[flavio81]

Most pro work in cinematography is still done with hand held metering so no TTL, (...)

I think you've hit the nail on the head. Typical 35mm SLR shooters use TTL metering, so any light loss due to the lens is compensated.

Especially with early zooms, with elements having very rudimentary anti flare coatings, flocking, baffles, etc. it was found that some zooms at some settings were only about 85% (average) efficient in getting light through the pipe. Lens coatings, etc. have really improved but for critical professional work time is still money so the less fiddling a cinematographer has to do the better.

+1
Zooms really suffer loss of light transmission with single-coatings.

 

[cowanw]

I am not sure that TTL solves the problem. TTL is tied to the maximum aperture of the lens, which must be communicated to the camera. That aperture is still measured as an F stop and is inaccurate as any other F stop.

 

[Luckless]

It would depend on how exactly the TTL metering is designed, but I'm not aware of any that really address the issue.

The general inaccuracy of TTL meters with 1/2-1/3 stop steps really doesn't help either. I haven't had my hands on much T-marked gear, but I was under the impression the accuracy was far tighter than normal SLR metering would account for.

You also have the risk of iris flutter on gear that stops down for the shot and meters wide open. You could build a accurate meter system that carefully stops down and adjusts to metering values, but any that I've seen appear to set an f/stop target and quickly jerk the aperture down to that with the risk of not actually hitting 100% the same mark each and every time. I had one lens on an EOS mount that seemed to struggle with this, with about a 1/10th of a stop spread near middle aperture ranges. - For the most part it really wasn't noticeable if you weren't trying to do a series of high speed shots.

 

[flavio81]

I am not sure that TTL solves the problem. TTL is tied to the maximum aperture of the lens, which must be communicated to the camera. That aperture is still measured as an F stop and is inaccurate as any other F stop.

F stop is accurate because it is geometrical. The max aperture is communicated accurately? depends on the system. For example on the Pentax-KA mount it is communicated in a very accurate way (digital).

After the camera knows the F-stop, it meters through the lens, so the light loss of the lens is accounted for. A different lens with different light loss will meter differently, which is what it should happen. So no problem.

 

[Luckless]

However that is not all that accurate or reliable in many lenses. Overall light transmission is not always perfectly even throughout the aperture range of a lens, as you stop down with simple geometric f/stops you may drift up or down from the expect overall transmission value. This issue is then further compounded with zooms - My Sigma 150-500 at the long end ends up being more than half a stop off from my canon 70-200 f/4L's exposure.

 

[flavio81]

However that is not all that accurate or reliable in many lenses. Overall light transmission is not always perfectly even throughout the aperture range of a lens, as you stop down with simple geometric f/stops you may drift up or down from the expect overall transmission value. This issue is then further compounded with zooms - My Sigma 150-500 at the long end ends up being more than half a stop off from my canon 70-200 f/4L's exposure.

This is true and depends on the lens. Some lenses have very good calibration on f-stops, others don't.

For all i love the Canon FD system, i've had to re-adjust the f-stop calibration on some of them, for it was off sometimes. But this might be because of a previous technician messing with the lens.

 

[wiltw]

I think you've hit the nail on the head. Typical 35mm SLR shooters use TTL metering, so any light loss due to the lens is compensated.

But since TTL SLRs and dSLRs almost all meter with lens wide open (the exceptions being the Spotmatic-like TTL M42 cameras from multiple manufacturer) any imprecision due to aperture blade closure error is not factored in, although the fundamental lens transmission (groups/elements, coatings) does get accounted for. Have there been any tables published about differences in light transmission between stated f/stop vs. the t/stop actual transmission at stopped down values?

F stop is accurate because it is geometrical. The max aperture is communicated accurately? depends on the system. For example on the Pentax-KA mount it is communicated in a very accurate way (digital).

But as lens tests by Popular Photography and by Modern Photography illustrated in articles from 50 years ago, the max aperture of the lens was NOT so accurately reflected in actual vs. engraved value!
The so-called f/2.0 lens might measure f/1.91, the f/2.8 lens might measure f/2.93, for example.

 

[flavio81]

But since TTL SLRs and dSLRs almost all meter with lens wide open (the exceptions being the Spotmatic-like TTL M42 cameras from multiple manufacturer) any imprecision due to aperture blade closure error is not factored in

Yes, this is what Luckless says above.

Old magazine tests (modern photo?) indicate the T-stop measured at wide opening.

 

[wiltw]

Yes, this is what Luckless says above.

Old magazine tests (modern photo?) indicate the T-stop measured at wide opening.

Don't think that T stops had anything to do with measurements! For example, Olympus 21mm f/2 lens measured actual: 20.15mm / 1.91
f/2 aperture at 21mm = 10.5mm diameter aperture; 20.15mm/10.5 = 1.9190476

 

[cowanw]

I remain unconvinced. If the camera is told that the amount of light transmitted at full aperture is (for example) 1.8 as an Fstop when it is actually 1.9 as a T stop, then the camera cannot take incorrect input and yield correct output.
My first camera was a Topcon UNI which had leaf shutters in the lenses. To maintain auto exposure I had to set a lever to the max aperture of the lens, to tell the camera what the perceived level of light meant (so it could choose a shutter speed). If I forgot to change the setting the camera erred, thinking the level of light might be 1.8, when, with a slower lens, it was actually 2.8.
The point I am making is if the Max aperture is in error (ie not a T stop) then the camera will be in error (although always proportionally changing according to its program, as the light level changes).

 

[flavio81]

I remain unconvinced. If the camera is told that the amount of light transmitted at full aperture is (for example) 1.8 as an Fstop when it is actually 1.9 as a T stop, then the camera cannot take incorrect input and yield correct output.

The fstop indicates the actual aperture. You're confusing aperture with light transmission.

T-stop is a measure of light transmission. Which is a combination of the aperture with the light transmission loss caused by the lens.

On a typical 35mm SLR, the camera's TTL photometer knows the f-stop and also measures the light that passes through the lens. So the meter value incorporates the light transmission loss caused by the lens.

 

[flavio81]

Don't think that T stops had anything to do with measurements! For example, Olympus 21mm f/2 lens measured actual: 20.15mm / 1.91
f/2 aperture at 21mm = 10.5mm diameter aperture; 20.15mm/10.5 = 1.9190476

I don't mean "mechanical measurements". I've seen magazine tests where, besides the actual measured APERTURE (actual measured f/stop in physical terms), they also compute the actual T-stop you get when the lens is wide open.

See:

Normal 50s 1977 by Jussi, en Flickr