Metering in a Canon 20D with Leica R lenses

[darcy_lorimer]

As I have mentioned in previous posts, I purchased a 20D as a

convenient, reasonable cost road to digital while still using my R

optics. While taking a number of test shots with the 20D body and my

Sumicron 90mm f2, I noticed very significant metering errors in the

20D. At wide open apertures, I was under-exposed by 1 to 1.5 stops

with the Leica glass. I confirmed this by shooting a grey stucco wall

in natural sunlight, and looking at the histogram which generally was

a sharp spike distribution due to the monochrome nature of the wall.

A Canon 100mm f2.8 AF lens gave a correct exposure with a centered

histogram distribution. My Leica f2 90mm gave a distribution shifted

significantly to the left at f2. As I stopped down, adjusting the

shutter speed to the exposure recomended by the 20D, the histogram

began moving to the right. It was centered at about f4 to f5.6, and

moved slightly to the right at higher stops to f22. I tried my 35mm

f2 Leica R lens and it did the same thing. The Canon lens gave a

centered distribution at all f/stop shutter speed combinations (but

you cannot meter with the Canon lens stopped down). I checked a

second 20D body at my local camera store and got similar results, so

I don't think my 20D is "deffective" (unless by design).

 

I understand that Canon meters at a fixed aperture and can alter the

exposures through ROM programming in the lenses. What is unclear is

why the Leica optics would put twice as much light into the finder at

the same f stop. For example, the Leica R lens at f2.8 was 1 stop

underexposed (more light to the meter) than the Canon at f2.8. I

guess I always figured that f2.8 is f2.8, no matter what the lens.

Unfortunately, this makes using the Leica optics complex because the

exposure correction is f stop dependent. Staying above f4 or f5.6

minimizes the error, but creates other problems.

 

If anyone has observed the same thing or understands why this is

occuring, I would be interested in your experiences.

[tim_franklin]

How exactly are you adjusting the aperture when the Leica glass is mounted?

 

You need to set the aperture on your Canon body to its widest value (either f/1 or f/0

depending on the model) and do all aperture changes on the lens. I initially has metering

problems with a Contax lens on the 1Ds I previously owned until I remembered about the

above.

[steve_unsworth1]

That mirrors my experience with the 5D.

 

I believe setting the aperture on the body is a custome feature missing from the 20D and 5D but included in the professional models.

[tim_franklin]

I don't think so Steve. I no longer have any Canon gear, so doubtless someone will correct

if I remember things wrong.

 

In manual exposure, the shutter speed and aperture are controlled separately by the two

dials on the camera (ie, the normal one in front of the release, and the big dial on the

back). In Av mode, the aperture selection is controlled from the first of those two (or by

the rear dial using custom settings), and the shutter speed is adjusted automatically

according to the changes in set aperture.

 

With a Leica lens mounted (or any other lens fitted via an adapter), closing the lens down

from its own aperture ring acts in just the same way as altering it on the Canon body - the

shutter speed will change automatically to compensate. Manual mode is similar, except for

having to change the shutter speed yourself.

[steve_unsworth1]

Tim, with an R lens fitted the displayed aperture is 00 in Av mode. Rotating the aperture dial doesn't change this.

 

If you shoot a plain subject - say a white wall - in Av mode, the image lightens as the aperture gets smaller. This is clearly seen as the histogram shifts to the right.

[bob_satroale]

The Leica might have oily blades. Send it to Sherry for a check up.

[steve_unsworth1]

Bob, I've tried the 5D with two 50s, a 180 and a 250 and had the same effect with all lenses. It would appear to be a function of how the camera's meter works - i.e. how it communicates with the lens.

 

I shoot in RAw so it's pretty easy to allow for the discrepancy during the conversion.

[nemeng]

D'Arcy - see the discussion on Canon EOS / manual lens metering errors in the Leica FAQ I maintain at:<p>

 

<a href="http://www.nemeng.com/leica/002f.shtml">

http://www.nemeng.com/leica/002f.shtml</a><p>

 

After the page loads, scroll down to the "Canon EOS snafus" heading.

[bobatkins]

(1) You can meter with a Canon lens stopped down, you just have to know how. You set the aperture, hold down the DOF preview button and remove the lens from the camera. It will be stopped down. You then put it back on the camera, but you don't rotate it to click into place, just enough to hold on the body. It then works like a stopped down, manual focus lens.

 

(2) Metering can be different because the position of the exit pupil of the Leica lens isn't necessarily in the same place as it is in Canon lenses and the Leica lens doen't have a ROM chip which tells the camera of any necessary exposure modifications. Exposure compensation when using non-EF series compatible lenses is common. They don't all need it and they don't all need it at all apertures, but it's normal. That's the way it works. Knowing EXACTLY why in every case isn't going to help!

 

(3) You'd be better off asking such question in the EOS forum where you might find people who know the answers.

[darcy_lorimer]

Thanks to all for their input. I made the measurements by puting the camera in manual mode, and yes, the displayed aperture is 00 with R optics. But the shutter speed is fully adjustable. Adjusting the aperture ring on the lens stops it down, and for each stopped down setting I adjusted the shutter speed until the exposure needle indicator was centered ("0").

 

So I proceded from f2 to f16 in this manner, took the images, and reviewed the histograms. The thing that caught my attention immediately was that the shutter speeds were not changing by the typical factor of two as I stopped down.

 

Now I can understand that Canon meters at a fixed aperture with their own lenses (manual or automatic settings), and tweaks the readings electronically for each lens. What I don't fully understand is why Leica optics would cause the exposure sensors to read so much more light at the same "reading" aperture. I've had this discussion with Rachael Katz of Katz Eye Optics, who has seen this phenomena with her special Canon focus screens at smaller apertures (below f4), and sees no difference with non-Canon lenses (such as Tamron and Pentax) wide open. It appears to be unique to Leica R optics, perhaps?

 

I also did another test, comparing the readings bewteen the Canon 20D w/ Canon 100mm f2.8 Macro lens and my Leica R7 w/90 mm f2. The subject was the same grey wall, same lighting conditions. Both read an exposure within 1/3 stop of each other.

 

I don't understand the comment regarding "oily" blades. How would this introduce more light to the exposure sensors at wide open aperture? If at, f2.8 or smaller apertures the blades were not closing far enough, I should see over-exposure down through the complete range (all histograms to the left of center), which I do not. The histograms move left to right, through center, in a uniform progression, incicating correct exposure at f4-f5.6, then slight underexposure f8-f16. Also remember that the images are taken at a fixed aperture with the Canon, meaning the blades don't have to stop down and open back up as they would on a Leica body. So there are no frictional or dynamic effects on the aperture blades.

 

The good news is that the exposures are not off by more than two stops. If you take images in RAW format, the Canon software utility allows up to a two stop adjustment. However, for complex scenes with dark shadow areas, you could still loose shadow detail information even with the correction.

[darcy_lorimer]

Andrew:

 

Thanks for the link. You website is very helpful. It does seem, after reviewing the references you provide, that this overexposure affect is seen primarily with Leica and perhaps Zeiss lenses.

[darcy_lorimer]

Bob Atkins:

 

(1) This technique is not intuitive or described in the manual. What results do you get when you use it? Since the lens is physically disconnected, no ROM or electronic info can go to the body. Are the same errors seen?

 

(2) Yes, the discussion is somewhat academic. The convenience of digital does allow multiple exposures to made quickly and analized with the histograms for correct exposure, as long as you have the time and your lighting conditions aren't rapidly changing. But the problem does reduce the attractiveness of using R optics on the 20D, in addition to the inconvenience of having to use the lens stopped down at the working aperture.

 

(3) I will post something in the EOS forums, but I posted here because the problem (ie underexposure at large apertures) seems to be unique to Leica (and perhaps Zeiss) lenses. And it was on this forum were folks suggested using the 20D for digital with Leica R lenses in the first place. So I assumed that those using R lenses with the 20D would have observed this.

[jan_brittenson]

Stop-down metering only works with some Canon bodies, depending on which type of metering it uses. It didn't work at all with my Canon 1D, but worked fine with the 1Ds and 1Ds2 for instance. I used these with both Contax and Leica R lenses, with no difference in results. I suspect the 20D's meter needs to know what aperture it's metering at, or it might be close to the edge of the image so needs calibration (linearization) data from the lens.

[joseph_wisniewski]

The Chuck Westfall explanation on Andrew's site is a step in the direction of the truth (Bob's theory about exit pupil location is a fascinating side trip, but doesn't take you anywhere that you need to go).

 

Modern cameras have focusing screens that don't scatter light as much as the screens of a couple of decades ago. Because the screen doesn't scatter much light, the metering system is seeing mostly rays that pass through the aperture close to the center ("near chief" rays. The chief ray is the one ray that passes directly through aperture center). That's also true of the light that passes through the prism to your eyes, and is the reason that you see virtually no change in viewfinder brightness stopping down from f1.4 to f2, and very little going from f2 to f2.8. As you get to smaller and smaller apertures, you get to the point where you (and the metering system) actually see a one stop change for evert stop on the lens. For modern focusing screens, this point is about 10 degrees, or f5.6. Any aperture wider than that needs to have it's readings compensated a bit, or the camera (reading mostly just the center portion of the light cone) thinks there isn't as much light coming through the lens as there really is, and overexposes, in an attempt to compensate.

 

If the camera knows the aperture of the lens, it can apply compensation automatically via the "program curve" that Westfall referred to. If not, the camera just assumes some "safe" aperture, probably around f4 (since large stops overexpose, small ones underexpose).

 

If you want to "fix" this, find the KatzEye notes on how to change the screen in a 20D, get a spare screen from Canon (a KatzEye is a bit too much money to mutilate), and rough it up with 360 grit carbide powder to make an old fashioned focusing screen. It will then be off by a constant amount, no matter what aperture you use, so you can easily learn to set the exposure compensation to one value that will "fix" all your Leica lenses, no matter what aperture you set them to. Do not rough up the wrong side of the screen ;)

 

You'll probably find the rougher screen easier to focus manually than the origional, brighter screen. The origional screen has too much similarity to an aerial image, not as much DOF as it should have at wide apertures, so you don't see things "snap" into focus.

[nemeng]

<i>that this overexposure affect is seen primarily with Leica and perhaps Zeiss lenses</i><p>

 

Not so. AFAIK it applies to all lenses which don't have the Canon EOS chip.<p>

 

I suspect the Leica / Zeiss crowd complain about it more because they're more likely to shoehorn their expensive glass onto different manufacturer's bodies :?)

[matt_sallis]

I have used a wide range of Zeiss C/Y lenses from 21mm to 300mm on a 20D for 16 months, and more recently on a 5D, and my experience reflects yours. I have also heard this from many others Contax and Leica users who are using Canon digital bodies on their lenses.

 

The exact exposure innacuracy varies from lens to lens, but seems to follow a similar pattern, ie. overexposed at wide apertures and underexposed at smaller apertures.

I don't really understand quite why it happens, and I don't think there is any solution other than using exposure compensation.

 

Depending on the number of lenses you use, you may be able to remember just what compensation is required, and dial this in whilst setting the aperture. Generally I follow the 'decrease wide' and 'increase stopped down' pattern, and it soon becomes second nature. One thing I have noticed is that matrix metering tends to be the most accurate.

 

Regards,

Matt

[darcy_lorimer]

Andrew:

 

The reason I cite Leica/Zeiss lenses is that the phenomena has been mainly reported by owbers of 20Ds and these lenses, and my detailed and lengthy conversations with Rachael Katz have led me to believe that a number of other lens brands do not show the under-exposure effect at wide open apertures. Rachael has personally tested a number of 20Ds sent to her for screen changes, and has never observed the under-exposure effect wide open with Tamron or Pentax lenses. She has observed mainly a requirement to compensate for over-exposure at small apertures. So I don't think this has anything to do with Leica lens owners reactions, its a real affect with Leitz lenses.