Sigma 50mm Art underexposing

[nemophoto]

I was wondering if anyone else who owns this lens has noticed this. First of all, I really like the lens in terms of handling and sharpness. It's superb. I sold my Canon 50/1.2, because I had to stop down to 5.6 to get decent sharpness across the field. Sort of defeats the purpose of a fast lens. But that's another story (and post elsewhere here).

I've owned the lens for about 9 months. On Av, I have no issues with the lens -- it works well. BUT, on manual (which comprises 98% of all my commercial work -- on location and in studio), I've discovered that my images are about 1/3 to even 1/2 stop darker than any of my Canon lenses. So, if my exposure is, let's say, f4 at 1/250, I have to open the lens or the shutter speed another 1/3 or so. Granted, not a huge issue, but bothersome and one that I have to consciously remember to adjust. At first, I didn't realize that this was happening and chalked it up to lighting changing subtly or my screw up.

I sent the lens to Sigma. It checks out fine on their equipment. They don't (needless to say) have access to Canon lens to check what I'm saying, though they accept, without an explanation, that that may be happening. I hate to get rid of the lens, but may do that depending upon what Canon releases.

Anyone else run into this? When I get the lens back, I may post some comparative screen shots.

[AlterEgo]

I've discovered that my images are about 1/3 to even 1/2 stop darker than any of my Canon lenses. So, if my exposure is, let's say, f4 at 1/250, I have to open the lens or the shutter speed another 1/3 or so.

how did you test ? with rawdigger to see/compare sensor saturation with equal nominal parameters of exposure and did you see/compare T-stops tests (for example by DxOMark)... Sigma 50/1.4 Art is T1.7 stops ( http://www.dxomark.com/Lenses/Sigma/Sigma-50mm-F14-DG-HSM-A-Canon-mounted-on-Canon-EOS-5D-Mark-III---Measurements__795 ) ... what about your other Canon lenses ?

[nemophoto]

It wasn't hard to compare. I could see the difference on shots taken with the Sigma 50 versus the Canon lenses. I know (having worked in cinematography years ago) that there is a difference between T-stops and f-stops (a mathematically derived aperture versus actual light transmission). But, the difference was seen, not wide open, but at anything from f3.5 to 5.6 on my 1Dx and my 5Ds. To be honest, the 1Dx has a little higher sensitivity than my 5Ds (hence brighter by a little), but this is comparing Canon lenses to the Sigma on the same body. Once I get the lens back from Sigma, it will be pretty easy to do a comparative test.

[AlterEgo]

It wasn't hard to compare.

compare where exactly ? you need to use the software that shows how data is in raw channels before any hidden corrections done by DPP or whatever  ... plus T stop vs F stop is different at any aperture wide open or not, it is not 100% about vignetting...

[nemophoto]

Compare how? Easy. With the Mark I eyeball. Still an effective tool. I may not be able to see the "numbers", but I can certainly perceive light and dark. I suppose the advantage of having been a photographer for 40-years. Of course it's not about vignetting. It's about the overall exposure variation. Anyway, I suppose yet another new 50mm is in my future. I'll wait for the new Canon 50 1.4 that I'm sure is on the horizon. It's just disappointing since the Sigma is a good lens in many other ways -- just not a match, exposure wise, to my other Canon lenses.

[Jim Pascoe]

I have no explanation, but I would have thought it not a major problem to remember to allow for this fact when exposing.

Just one thought - if comparing to the Canon lens, would the actual length of the lens have any bearing on the issue.  I remember when shooting with cameras that used bellows focussing one had to allow for light fall-off as the lens was racked out.  If the Sigma is significantly longer than the Canon would the effect not be to reduce the amount of light reaching the sensor?  Or has that already been accounted for in calculating the f-stop.

I expect an expert here will be able to comment.

Jim

[Bart_van_der_Wolf]

I have no explanation, but I would have thought it not a major problem to remember to allow for this fact when exposing.

Just one thought - if comparing to the Canon lens, would the actual length of the lens have any bearing on the issue.  I remember when shooting with cameras that used bellows focussing one had to allow for light fall-off as the lens was racked out.  If the Sigma is significantly longer than the Canon would the effect not be to reduce the amount of light reaching the sensor?  Or has that already been accounted for in calculating the f-stop.

I expect an expert here will be able to comment.

Hi Jim,

I'm not an expert, but it is a strange thing and not related to the so-called "Bellows factor" (purely a magnification factor issue). The F-number is in principle given for infinity focus, the effective aperture becomes narrower (the  F-number higher) at all closer focusing distances. It's a purely geometrical ratio number, focal length divided by (apparent or effective) diameter of the aperture, or f/#.

A T-stop is the F-stop, but now corrected for actual lens transmission, and it's the transmission which might cause the observed lower amount of exposure. It does seem a bit extreme though, so I'm not 100% confident that it is the real cause. Maybe all apertures are too narrow (poor adjustment?), or some of them are not closing correctly in a uniform progression?

Maybe the electronics report the wrong EXIF values and the Raw converter adjusts based on that? Many things to eliminate, but that requires shooting sequences of Raw files on manual exposure settings and judging the Raw data with e.g. Rawdigger, before conversion. If all other possible causes are eliminated, then a lower T-Stop transmission becomes more likely.

Cheers,
Bart

[Jack Hogan]

The F-number is in principle given for infinity focus, the effective aperture becomes narrower (the  F-number higher) at all closer focusing distances. It's a purely geometrical ratio number, focal length divided by (apparent or effective) diameter of the aperture, or f/#.

I don't mean to be pedantic, but many of the better formulas in optics cannot be easily understood (:-) unless one knows that f-number N is defined as

N = 1 / [2*sin(theta')]

with theta' half the opening angle of the focused light beams (Nakamura).  At infinity focus it can indeed be approximated by f/D.



Jack

[AlterEgo]

I don't mean to be pedantic

what about the effective aperture of microlenses ? when you mount f1.2 lens on a particular camera ?

[Jack Hogan]

what about the effective aperture of microlenses ? when you mount f1.2 lens on a particular camera ?

As far as I know N is purely a function of the optics, as shown in the simplified model in the picture above.

Jack