Question: 5D exposure metering readings in Av mode are different for 17-40L and 16-35L.

[sevtapisik]

Recently I've been shopping for a wide angle zoom lens. Yesterday I had a chance

to test both 17-40L and 16-35L on my 5D. I covered all f-stops at the focal

lengths of 16, 17, 24, 35 and 40mm.

 

There was one puzzling observation I could not make sense of it: In Av mode, all

17-40L shots were 1/3 to 2/3 stop darker than corresponding 16-35L shots.

Meaning, under the same settings/lighting conditions, the camera metering was

consistently different for the lenses. When I shot manual, the outputs were

close enough to be called identical. In case you wonder, the camera metering was

evaluative in Av mode.

 

I could not come up with an explanation for it and I am quite curious. Any ideas

about what is happening?

 

Thinking that they are not relevant, I am not including the general results of

my tests in this thread. I will be more than happy to share them with anyone

interested in. Also I have all the raw test files if you'd like to see and make

your own conclusions. Feel free to email to me.

 

Thanks!

[pjmeade]

That's an interesting observation. I don't keep much of a watch on these things, but I don't remember anyone reporting that before. One would expect that at a given f-number, shutter speeds would be equal and that the exposure would be identical. I'm reassured that exposures were the same when you used manual.

[michael.gregory]

I don't find this at all problematic. You say that in Manual mode that the readings were "close enough to be called identical". While I don't claim to actually "know" the reason why, I would guess that the Analog to digital converter used to measure light intensity must have a quantization threshold which is established in firmware. In other words some engineer for canon wrote the code so that a 1 bit difference in the measurement taken from thid A/D converter means some number of lumens, (X). Very likely there are a number of these quatization factors established for any number of different metering conditions and user selected exposure quantizations, (1/3 vs 1/2 stop).

It may well be that in manual mode the program requires a difference of 2 bits from the A/D converter to cross a threshold of 1/3 stop, while in an automatic mode the program only requires a one bit difference for a step. the thresholds might well have been established differently because the code was written by different programmers, or for some more esoteric reason, but in the larger view, so long as the metering is correct within 1/2 stop, and consistent with the same lens in the same mode under the same conditions, I would not think that it much mattered, Calibrating this out to the individual taste of the user is what exposer offset is all about after all.

[HenryUK]

The metering sensor in the camera measures a certain intensity of light falling on it. In order for the body to calculate the exposure it needs to know what aperture the lens is at to determine the actual illumination of the subject.

 

The lens transmits its maximum aperture to the body, so that the body knows how to interpret the metering sensor readings.

 

If the actual maximum aperture is larger than the reported value, the body will think the subject is lighter than it really is, and underexpose.

 

There are tolerances for the parameters of the lens. I could believe that the maximum aperture is only within plus/minus a third of a stop of the nominal value. Two thirds of a stop error sounds too much.

 

Do you get the same error in spot metering mode? It's possible that the two lenses have different light distribution patterns, e.g. vignetting, which is fooling the evaluative metering.

 

Henry

[baivab]

From 2 responses - it looks like a bug, than an explanation. My logic being, whatever said - a light meter takes all things into consideration and responds with a F-stop against a shutter speed or vice-versa or all together. The exposure should be consistent. Period. Do you have ANY other lens - primes even. Can you verify whether the other lens matches either one of the lens you've tested?

[m_barbu1]

<p><i>Meaning, under the same settings/lighting conditions</i></p><p>Out of curiosity, what were the lighting conditions? What were (and where were) your light sources?</p>

[sevtapisik]

Henry Clark: I did not do spot metering and 16-35 is gone now. Your suggestion about the different light distribution patterns fooling the evaluative metering makes sense to me.

 

Baivab Mitra: I do have a 24mm prime. Since I ended up keeping the 17-40, I will test them against each other.

 

M Barbu: Your questioning the lighting conditions is fair enough. I did the testing indoors, with window light, no direct sun in view on a clear day. I have no way of detecting 1/3 to 2/3 of intensity change without actually metering for every shot. But the compared images were taken in the same 20 min. time frame -approximately- and not in a particular order. Meaning I kept switching between the lenses many times and my testing was completed in two days. So when I claimed that the lighting conditions were the same, I was convinced that I ruled out that possibility. My observation was consistent throughout the process and also note that the manual exposure results were identical.

 

Thank you all for your valuable inputs. I was more curious than concerned about this issue.

[m_barbu1]

I have found that the intensity of the light from the sun varies as various objects pass inbetween the sun and my subject, and as the sun's position changes in our sky. I would not think it strange to have such a fluctuation if you were not relying on a scientifically known constant source of light. You might wish to repeat the test in a room with no windows (or at night), using an incandescent light bulb as your light source, and ensure that the camera and light are in fixed places (and that nothing else is moving).