[Robot camera]

I have found that a couple of my manual focus lenses do not produce the same exposure levels as others. Today I tried four lenses, all on manual exposure, set to 1/45th @ f/5.6. These were the results, with the exposure from my 16-50 Pentax lens as '0'. That is, correct. My camera is a K20D. The target was an evenly lit magnoli painted woodchip papered wall. It's a quick and dirty test, but nevertheless there's a problem. Going by the histogram, I found:

Pentax 16-50mm f/2.8 @ 50mm: 0
Tamron 80-210 f/3.8-4 @ 80mm, KA mount: -0.2 stops
50mm f/1.4 Pentax-A: -0.8 stops
24mm f/2.5 Tamron, KA mount: -0.8 stops

Strangely, out in the real world the 24mm Tamron over exposes by half a stop, The camera gives consistent and accurate exposure with my 10-17, 12-24, and 16-50 Pentax lenses and a 28-105mm Sigma, though I have found that permanently leaving the exposure compensation on plus half a stop gives me exposures that I like. Has anybody got any idea why I have this problem with the 24 and 50mm lenses? Is there a cure?

[Sagitta]

Not every lens is or ever was manufactured exactly the same. I'd bet if you picked up three other 16-50's, all four would have some degree of variance as well.

I'm actually not seeing where this is a problem at all, considering its a one-button (or one aperture-turn) fix when swapping out lenses.

[UncleVanya]

The a series Pentax lens coatings aren't the same as the 16-50- there might be more transmission losses.

[enoeske]

I have found that a couple of my manual focus lenses do not produce the same exposure levels as others. Today I tried four lenses, all on manual exposure, set to 1/45th @ f/5.6. These were the results, with the exposure from my 16-50 Pentax lens as '0'. That is, correct. My camera is a K20D. The target was an evenly lit magnoli painted woodchip papered wall. It's a quick and dirty test, but nevertheless there's a problem. Going by the histogram, I found:

Pentax 16-50mm f/2.8 @ 50mm: 0
Tamron 80-210 f/3.8-4 @ 80mm, KA mount: -0.2 stops
50mm f/1.4 Pentax-A: -0.8 stops
24mm f/2.5 Tamron, KA mount: -0.8 stops

Strangely, out in the real world the 24mm Tamron over exposes by half a stop, The camera gives consistent and accurate exposure with my 10-17, 12-24, and 16-50 Pentax lenses and a 28-105mm Sigma, though I have found that permanently leaving the exposure compensation on plus half a stop gives me exposures that I like. Has anybody got any idea why I have this problem with the 24 and 50mm lenses? Is there a cure?

The hypothetical aperture (or ideal aperture, what it was meant to be) is different than the effective, actual aperture. This is called the T-stop.

[UncleVanya]

The hypothetical aperture (or ideal aperture, what it was meant to be) is different than the effective, actual aperture. This is called the T-stop.

True, my perception was that zooms often have a higher T-Stop value (vs. their F-Stop) than primes. My guess however is that this difference is very small and that the older lens coatings make more of a difference than the number and arrangement of elements on the actual light transmitted to the sensor.

[jatrax]

Older lenses often expose differently on modern DSLRs. It could be difference in the coatings or internal design of the baffles or because the camera gets no information from the lens since they have no ID chip. When using an older lens I usually take a test shot and adjust exposure comp to that lens.

[Adam]

Older lenses often require exposure compensation. My A and K 15mms always underexpose by around half a stop, for instance.

[Robot camera]

Older lenses often require exposure compensation. My A and K 15mms always underexpose by around half a stop, for instance.

I've just done another quick test. In aperture priority, with centre weighted exposure, there is the same variance. So it is not a difference in the light transmission of the lens, as this would be compensated for by the metering. Also, I cannot see that there is as wide a production tolerance as nigh on a full stop! What jatrax and Adam have said is of interest, although I have never heard of this problem before, and I have been around a long time. It's not that I don't believe you, but surprise that nobody says anything! People talk about the backwards compatibility advantage on the 'net all the time, but nary a word about this! It's not as if they were K or M mount lenses. That I could understand.

[jatrax]

What jatrax and Adam have said is of interest, although I have never heard of this problem before, and I have been around a long time.

I don't think it gets much mention because it is not a problem. You just set the exposure comp per the lens used. In most situations the variability in the scene being metered will exceed the variability introduced by the lens so it goes unnoticed. The only time you will see it is when you do controlled tests. I did a complete test on about 15 Takumar lenses a few years ago. Using Av mode and attempting to determine if metering accuracy was consistent across different apertures. The results indicated no discernible pattern across all lenses except that with those lenses metering varied as much as 1.5 stops across the range of aperture settings. At some apertures it was perfect and at others varied either +/- and different lenses showed different results.

[UncleVanya]

On the plus side, with digital you know if it was exposed close as soon as you take the time to look.

[StephenHampshire]

I don't think it gets much mention because it is not a problem. You just set the exposure comp per the lens used. In most situations the variability in the scene being metered will exceed the variability introduced by the lens so it goes unnoticed. The only time you will see it is when you do controlled tests. I did a complete test on about 15 Takumar lenses a few years ago. Using Av mode and attempting to determine if metering accuracy was consistent across different apertures. The results indicated no discernible pattern across all lenses except that with those lenses metering varied as much as 1.5 stops across the range of aperture settings. At some apertures it was perfect and at others varied either +/- and different lenses showed different results.

None of my lenses meter identically, whilst most of my modern (DA) Pentax lenses are pretty similar, my Sigma 50-150 f2.8 needs -0.5 EV to be "correct to my taste.
Also, on lenses where the camera meters wide open, it has to apply a certain amount of movement to the lens aperture lever to stop down to what it thinks is the correct f-stop - if the calibration of the lens aperture lever mechanism is out,this can result in the lens stopping down more or less than the value than camera thinks it is setting, which will lead to inconsistent results. Generally, I know what EV I should use for my most popular lenses, none of mine are out by more than 0.5 EV

[Lowell Goudge]

Years ago now, when exposure metering was a big topic, I ran a ton of tests, using my *istD and K10D. I have since continued on with my K7D and K5D

Exposure carries between bodies with manual lenses, and between lenses on the same body.

Even adding a teleconverter changed the exposure of my sigma 70-200/2.8

What I found is that the exposure with manual lenses is quite non linear as a function of the camera and focusing screen and the actual aperture you meter with. For A lenses, the camera is largely compensated with the aperture knowledge and very close to correct.

You should test every lens with every body, taking a series of shots using a uniform surface such as a paved road or block wall.

Even modern lenses like my Tamron 28-75/2.8 have some irregularities, it is perfect wide open and gradually exposes more and more when stopped down until it is one stop over when fully stopped down.

Shooting a series of shots to validate exposure on each lens is a worthwhile 10 minutes per lens spent to know your gear.

[jatrax]

whilst most of my modern (DA) Pentax lenses are pretty similar,

I wonder if DA lenses have info in either the lens chip or the camera that allows more precise metering by describing the characteristics of the lens?

What I found is that the exposure with manual lenses is quite non linear as a function of the camera and focusing screen and the actual aperture you meter with.

That agrees with what I found in my tests. I expected a linear error based on static parameters but all I could determine with accuracy was that there were errors in metering that varied with the aperture. It might be possible to chart those errors and determine something but I never carried it that far, it certainly was not consistent. My tests were all on Takumars so the aperture linkage was not a factor.

[Lowell Goudge]

I wonder if DA lenses have info in either the lens chip or the camera that allows more precise metering by describing the characteristics of the lens?

That agrees with what I found in my tests. I expected a linear error based on static parameters but all I could determine with accuracy was that there were errors in metering that varied with the aperture. It might be possible to chart those errors and determine something but I never carried it that far, it certainly was not consistent. My tests were all on Takumars so the aperture linkage was not a factor.

Been there, done that,



Systematically it is based upon the apterature but there is no linearity, just consistency

[jatrax]

Systematically it is based upon the apterature but there is no linearity, just consistency

Yep, looks quite like the spreadsheet chart I came up with. I suppose there might be some mathematical formula to describe the curves but that is beyond me. Consistent and repeatable but no linearity.