5D with 50 f1.8 under exposure issue

My 5D with the cheapo 50mm f1.8 under exposes by one full stop at the largest

aperture ONLY - once I close the lens down by one stop, the exposure is dead

on. I figured maybe there is an issue with the cheapo lens, so I mounted it on

my 40D and the metering was spot on for all apertures.

 

I don't have any other fast lenses to do any further tests and I am dumb

founded.

 

Does anyone have any explanation for this?

Is it under-exposing, or just vignetting?

Colin, it is clearly under-exposing. The histogram confirms it.

If you take one shoot with f/1.8 and another one with f/2.8 ... does exposure time change as expected?

Rainer, no and that's the issue. When the exposure is 1/250 at f2.8, it jumps to 1/800 at f1.8.

vignetting would give a "darker" histogram even though the shot is still exposed the same...

Why would a 50mm on a full frame camera vignette? Could someone with the same setup - 5D anmd 50mm - do a quick test to see if the outcome is the same?

I used an EF 50 1.8 (MK I) on my 5D for a while. It had absolutely no vignetting. If you

have vignetting you may be using too large a hood or filter(s). However my EF 50 1.8 did

exhibit moderately severe light fall-off in the corners (same as it did when I shot slides).

Didn't normally matter as I only used F1.8 in murky scenes where the edges were normally

dark anyway.

 

As for exposure, normally I found the camera tended to overexpose dim scenes and

normally I need to pull back .5 to 1 stop to maintain the natural look. Most dim landscapes

have lack highlights so the histogram should be bunched up in the middle (unless you

practice exposing to the right and pull everything down a stop before RAW conversion).

Just tested the same 50mm lens on my Elan 7e film camera and the exposure meter behaves exactly like the one in the 5D - in other words it would also under-expose at the largest aperture. I do not use any filter or lens hood on that lens for these tests. I guess I just have to make a note to adjust exposure by +1 when shooting wide open. Interesting and curious never-the-less.

 

If someone with the same setup would be kind enough to test the same scenario, I would be very grateful.

Juergen, the reason a 50 on a full frame mighb vignette as opposed to on a crop sensor is that the crop sensor is only utilizing the "sweets spot" center of the glass. It's the edges of the lenses that are more inclined to vignette and the full frame sensor needs all the glass real estate to cover it's larger size. However, that doesn't seem to be your problem. 1/250 F2.8 should be 1/500 F1.8 not 1/800. The only thing I can think of is a communicaiton probem between the camera and lens. Perhaps if you cleaned the contacts, especially on the 5D since it worked on the 40D. Or I suppose it could just be a coincidence that the aperture sticks a bit and only manifested on the 5D not the 40D.

Your 5D obviously cares about who and what it hangs with :-)

 

Rainer T. asked a pertinent question, but the numbers don't quite jibe:

 

"50mm f1.8 under exposes by one full stop"

 

"When the exposure is 1/250 at f2.8, it jumps to 1/800 at f1.8."

 

Latter exposure is actually only 1/3 stop faster than 1/250 at f2.8.

Bruce, yes, it is not quite one stop, but close to 2/3 rather than 1/3. The correct exposure should be 1/500 - instead it jumps to 1/800, which is more than 1/2 stop. Bob, when I saw this under-exposure I immediately thought of dirty contacts and cleaned them, but that did not make any difference. And now my old film camera behaves the same way - so I'm thinking it is the lens!

Juergen,

 

f1.8 is 1/3 (or maybe 1/4) of a stop above f2, which is one full stop above f2.8. If I've got this right, f2.8 @ 1/250 = f2 @ 1/500 = f1.8 @ 1/640. That would make f1.8 @ 1/800 about 1/3 a stop of relative overexposure - no?

I got a bit curious, so I had to try this, so I made a series of equal exposures in 1/3 stops from f1.8 up to f2.8, and indeed, even though the (theoretical) exposure was exactly the same, each third of a stop gave a gradually lighter result. I then bracketed the exposure times downwards from f2.8 to compare to the f1.8 shot, and the result from f1.8 looks like it's 1/3 stop darker - that is disregarding the vignetting.

So, with the correct calculations, it looks like my result is the same as Jürgen's - a third of a stop darker at f1.8 than at f2.8. A third of a stop isn't any problem, though, since it makes only a minor adjustment in post-processing to equalise exposures...

Blushing - yes, Bruce you are right. It is still amazing how much 1/3 of under exposure shows - the picture does look considerably darker than just 1/3 stop. Anyhow, the mystery (that wasn't) is solved.

 

Thanks everyone for your help!

"each third of a stop gave a gradually lighter result"

 

Hakon,

 

I've noticed this too on my 350D -- exposures which in principle are equivalent sometimes yield noticeable differences in practice. I attribute it to the dSLR equivalent of rounding off error, but that's just because I'm reluctant to invoke the notion of the gremlins, which I think are the real cause :-)

 

 

Juergen,

 

I agree the mystery isn't totally solved, but I think Hakon's experiment might provide the nub of an explanation.

As you will notice, J�rgen, in addition to the 1/3 stops, the corners are a lot darker due to the light falloff, and those are very big dark corners at f1.8. Easily corrected in post-processing, of course, but still.<p>

Whereas the main part of the image is 1/3 stop darker, the corners are indeed 1 full stop darker on the f1.8 picture.

I don't usually think a picture is severely underexposed before it gets past a stop or two, all depending on whether post-processing is able to deal with it without making too much noise. Then again, I like deep shadows with just a hint of detail at most, but I dislike blown highlights, so I tend to dial in a bit of underexposure in any case, all depending on the scene in question, starting with 1/3 to 2/3 stops for outdoor shots on sunny days to 2 stops for night shots - that is to make them still look like night shots, of course.<p>

Indoors and pictures entirely in the shade are the only times I don't adjust the exposure down.<p>

"Correct exposure" is, after all, a matter of taste and artistic expression. It would of course have been nice if our lenses didn't add a bit of its own creativity to the equation, which can make control freaks (I can imagine there must be several of those with cameras out there) freak out... =P

<<< It would of course have been nice if our lenses didn't add a bit of its own creativity to the equation >>>

 

Yes understood, but (all) prime lenses will vignette wide open when used on the format camera for which they are designed. This is true of some zoom lenses too, especially the fast ones, and especially at the wide end of the zoom.

 

Faster primes (usually) show more degree of vignette, than slower ones: but very expensive (read `ace` quality) may be an exception to this generality: I expect that EF 50mmF1.2L would have less vignette at F1.2 than the 50mmF1.8MkII at F1.8, and this is due to the different lens design as well.

 

This is a fact of lens design in the practical world, as opposed to what we would like it to be in the theoretical. We just need to know it and adjust accordingly.

 

WW

William W wrote:

 

"I expect that EF 50mmF1.2L would have less vignette at F1.2 than the 50mmF1.8MkII at F1.8, and this is due to the different lens design as well."

 

According to SLRGear the vignetting of the f1.2 vignettes more at f1.2 than the f1.8 at f1.8. Indeed the f1.2 is literally off the scale with the corners 1.7 stops darker than the center. The f1.8 has corners 1.25 stops darker.

I did not know that (obviously) and thanks for pointing it out, I was lazy and did not check any lens test before offering that `opinion`.

 

These facts are just more to the points I was making about fast lenses, wide open and vignettes.

 

Thank you for addressing my laziness, Alistair.

 

WW

-- "Why would a 50mm on a full frame camera vignette?"

 

Because to an extend ... every lens shows vignetting ... this link ...

 

http://www.the-digital-picture.com/Reviews/Lens-Vignetting-Test-Results.aspx?Lens=105&desc=Canon-EF-50mm-f/1.8-II-Lens-Vignetting

 

 

reports the 50/1.8 to have 3 to 3.5 stops vignetting in the corner (wide open) ... and only about 1.5 stops vignetting when stopped down to f/2.8.

 

Since the histogram does value corners in the same way as the center ... but metering does not (at least not necessarily), the histogram for f/2.8 will be more on the right side than with f/1.8 (since metering was done with the lens wide open).

Why...would you put a hundred dollar lens on a 5D? And then wonder about quality!

Anthony, I have used the 50 1.8 on my 5D when the situation called for a light weight low light setup and I was surprised by the quality. Price is not necessarily an indication of the abilities of a lens... That being said, could it be that the lens is reporting its aperture as 1.4 when the camera is calculating the exposure? I'm not sure what goes on in the camera's little brain, but something like that might account for the subtle underexposure.