[Edgar_in_Indy]

This would not surprise me, really. Almost every lens I have sucks in different amounts of light at a given aperture.

I'm sure this is probably true, but I think that what I am seeing is much more than can be explained by differences in glass. Especially when one considers that the 17-50mm is supposed to be based on the same basic optical formula as the 28-75mm.

I did see tiny, *TINY* differences in exposures between the 28-75mm and my kit 18-55mm, but they weren't noteworthy at all. At some apertures the 28-75mm looked a hair brighter, and at other apertures the 18-55 was almost imperceptibly brighter. But the 17-50mm was noticeably and consistently darker than both.

If the darker image WAS caused by light lost in the elements, I think that would be even more disturbing than an incorrectly reported aperture, since it would completely negate the advantage of having f/2.8. But I don't think that's the case. I suspect that the aperture is slightly smaller than is being reported (at all apertures), and as the total size of the aperture decreases, the effect becomes more noticeable as the relative size difference increases.

I opened the three shots taken at f/14 in Photoshop in order to see how much I would have to brighten the 17-50mm to match the exposure of the other two. I had to move the exposure slider a full +EV stop in order to match the exposure of the other two lenses (but of course the adjusted 17-50mm image displayed more noise throughout the image).

Based on my findings I will be returning the Tamron 17-50mm. I usually shoot in fully manual mode, and I'm not going to put up with a lens where I have to remember to give it an extra stop comapred to my other lenses.

I'll probably stick with the 28-75mm as my primary lens, and add the Sigma 50-150mm f/2.8 to pick up where the Tamron leaves off, and then just keep the 18-55mm kit lens for the occasional instance where I need a wide angle.

A little further down the road I might pick up the new Sigma 17-50mm f/2.8 OS HSM. I've been looking at reviews of the new Sigma, and it looks they've got a real winner on their hands. Or I may just skip the 17-50 range all together and just add a Sigma 10-20mm (or something similar) to my kit.

[conradj]

That jives with what i'm accustomed to doing, that is, between f11 and f16 i get to +1EV compensation.

Guess i'm just used to it. A lot of my lenses need some kind of exposure compensation. It's too bad one can't program that into the camera the way one programs focal length into it for the SR.

[vicentesagredo]

straight from wikipedia:

T-stops
Since all lenses absorb some portion of the light passing through them (particularly zoom lenses containing many elements), T-stops are sometimes used instead of f-stops for exposure purposes, especially for motion picture camera lenses. The practice became popular in cinematographic usage before the advent of zoom lenses, where fixed focal length lenses were calibrated to T-stops: This allowed the turret-mounted lenses to be changed without affecting the overall scene brightness. Lenses were bench-tested individually for actual light transmission and assigned T stops accordingly (The T in T-stop stands for transmission),[3] but modern cinematographic lenses now usually tend to be factory-calibrated in T-stops. T-stops measure the amount of light transmitted through the lens in practice (actually on T-stops the amount of light is measured at the film plane), and are equivalent in light transmission to the f-stop of an ideal lens with 100% transmission. Since all lenses absorb some quantity of light, the T-number of any given aperture on a lens will always be greater than the f-number. In recent years, advances in lens technology and film exposure latitude have reduced the importance of t-stop values. So, F-stops are for focal ratio, T-stops are for transmission.

that's it.

[Edgar_in_Indy]

Well, it would be nice if T-stops were included in lens specs then. Although I suspect there's something more going on here than just the incidental light lost in transmission through the lens. Especially since the optics of the Tamron 17-50mm and 28-75mm are supposed to be closely related.

Out of the lenses I've owned for my K-x (Pentax DA L 18-55mm, Pentax DA 55-300mm, Sigma 18-125mm, Tamron 28-75mm, and now the Tamron 17-50mm), the Tamron 17-50mm is the first one where I've noticed such a large difference in exposure. I think that the f-stops are not being correctly set in the lens and/or reported correctly to the camera.

---

Last edited by Edgar_in_Indy; 11-28-2010 at 11:12 PM.

[jolepp]

I tried the Tamron 17-50mm 1:2.8 too and the results look similar. The shots below have been done under a single 40W tungsten spot while it was dark outside so light should be constant. The camera was mounted on a tripod and stayed in the same position.

#1 : f=2.8, t=1/20s
#2 : f=4, t=1/10s
#3 : f=5.6, t=1/5s
#4 : f=8, t=0.4s
#5 : f=11, t=0.8s

The difference between #1 and #5 is quite close to 0.5 EV or that is how much I need to boost #5 in ufraw to get it to as get the white on the soccer ball to about the same (+1/3 is darker, +2/3 is brighter).

[JHD]

If I understand this correctly, this exposure 'problem' with the 17-50 only becomes an issue when shooting in manual, is that correct?

[jolepp]

I would think that the effect is there in all modes; it seems as if the aperture closes a bit more than it ought to towards smaller settings.

[Edgar_in_Indy]

If I understand this correctly, this exposure 'problem' with the 17-50 only becomes an issue when shooting in manual, is that correct?

No, but the only reason I discovered it is because I always shoot in manual. I was shooting first with the Tamron 28-75mm in my studio, and then switched to the 17-50mm. Everything else--strobe output, camera settings, the subject, my location--remained the same. That's why I was so surprised when I downloaded the images and saw that the 17-50mm images were all dark/underexposed.

People who shoot in metered modes (Auto, Av, Tv) may not notice the problem, or may dial in exposure compensation assuming that the camera just does not meter correctly with the lens. When reading user reviews of the lens, I have noticed that a lot of people mention using exposure compensation with this lens.

If my copy was typical, then as far as I'm concerned Tamron should never have marketed this as an f/2.8 lens.

[Jewelltrail]

I own both the 17-50mm 2.8 and the 28-75mm 2.8. I compared the 2. My results are different than the ones here: the 2 lenses perform similarly and are very close in equal exposures. There is a slight difference between the 2, with the 17-50 being just a little bit darker. However, what I notice most, especially wide open, is the 28-75 handles vignetting better--this is expected since the 28-75 is a Full-Frame lens.

[Edgar_in_Indy]

I own both the 17-50mm 2.8 and the 28-75mm 2.8. I compared the 2. My results are different than the ones here: the 2 lenses perform similarly and are very close in equal exposures. There is a slight difference between the 2, with the 17-50 being just a little bit darker.

It actually sounds like your results are EXACTLY like mine, and further confirms this flaw. The Tamron 17-50mm was only half a stop darker at f/2.8. This would constitute the "slight difference" that you saw, and half a stop puts it around f/3.3 or f/3.4. My other two f/2.8 lenses showed NO difference, and the 18-55mm was also identical to those two at equivalent f-stops.

And I was using my lens in the studio to do high-key photography. So I was shooting with a white background that was lighted to be slightly overexposed, thus producing a pure white background. When I switched to the 17-50mm, the difference was enough to bring the background back into normal exposure, so the difference was more striking than you would see in most situations. And I was shooting at f/9, where the difference in exposure was more like a full stop.

It should also be pointed out that your monitor can make a difference in how observable the differences are. On my desktop computer with a 24" LCD monitor the difference is more noticeable than on my Sony Vaio laptop. The difference is even less noticable on the cheap LCD monitors I use at work. I think that since my 24" LCD has a very high contrast ratio you are able to see a bigger difference between darks and lights than you are able to see on more standard monitors.

[Edgar_in_Indy]

And just to be clear, this isn't meant to be a total slam on the 17-50mm f/2.8, which I believe is a very good lens. As I already mentioned, my copy produced beautiful, sharp pictures. But I believe in truth in advertising, and I don't believe it's right to market a lens as an f/2.8 lens when it is actually an f/3.3 lens.

That being said, I would still say this is probably one of the best lenses out there for replacing the kit lens because of its reasonable price. Given the price, image quality, build quality, and constant aperture it is a great lens for the money. But if somebody has more to spend and really needs f/2.8, I would probably have to recommend the new Sigma 17-50mm f/2.8, which by all accounts is a great lens, albeit significantly more expensive than the Tamron.

[Jewelltrail]

Edgar in Indy:It actually sounds like your results are EXACTLY like mine, and further confirms this flaw. The Tamron 17-50mm was only half a stop darker at f/2.8. This would constitute the "slight difference" that you saw, and half a stop puts it around f/3.3 or f/3.4. My other two f/2.8 lenses showed NO difference, and the 18-55mm was also identical to those two at equivalent f-stops.

No, I do not think so. I know what half a stop looks like and I also notice slight exposure differences between all my lenses. What I am talking about is a very slight difference--nowhere near a half stop. The monitor has nothing to do with it, because monitors are consistent. What I mean is, even though your monitor could be brighter or darker than mine, of course, the increments of light change would still be consistent across the aperture range with whatever monitor you use. That is, a half stop increment in my monitor would be a half stop increment in my monitor--equal throughtout the range.

If I were you, i would have called Tamron, and asked them for their advice. BTW, I do not take this as "slamming" anything. This is what our forum is for.

Anyway, it is fair for me to say I have not done anything scientific yet, but perhaps I will. If so, I'll post the results here.

[nater]

But I believe in truth in advertising, and I don't believe it's right to market a lens as an f/2.8 lens when it is actually an f/3.3 lens.

F-numbers are just a ratio, specifically focal length divided by the diameter of the entrance pupil. F-numbers do not take into account any light absorption that happens as light passes through the lens material - there is always some additional light lost. Because of this, you can't rely on F-numbers to be an absolute indicator of the amount of light that a particular lens lets through, but rather a ballpark indicator.

As someone posted earlier, T-stops are an accurate indicator of light transmission, but you'll only find their use in the movie industry (yes it would be helpful if the consumer photo industry used them as well, but they don't).

Incidentally, this thread reminds me of a thread a few months back discussing someone's tests of the Cosina 55mm f/1.2 vs. the Pentax 50mm f/1.2 (the Pentax lets in more light, probably because of the SMC coatings).

[paperbag846]

I would like to see shots from both lenses at exactly 50mm at f2.8, f 8, and f16, with identical ISO, shutter speed, and controlled lighting.

I think that the metering was thrown off, instead of the lens itself being darker. The original shots are at different focal lengths/shutter speeds, so it's really hard to tell what is actually going on.

I would suspect that the 17-50 would be a tiny bit darker because wide angles require light to bend more before it hits the sensor... but not to the degree we see with the OP.

If my hypothesis is correct, though, there should be little to no difference at 50mm.

[Edgar_in_Indy]

I would like to see shots from both lenses at exactly 50mm at f2.8, f 8, and f16, with identical ISO, shutter speed, and controlled lighting.

I think that the metering was thrown off, instead of the lens itself being darker. The original shots are at different focal lengths/shutter speeds, so it's really hard to tell what is actually going on.

I always shoot manually, so no metering was involved. I had started the shoot with the Tarmon 28-75mm f/2.8, adjusted my light and camera settings for proper exposure, and then started shooting. Midway through the shoot I changed to the 17-50mm f/2.8, leaving everything else identical. The 17-50mm shots were at around f9 or f10 were underexposed by a full stop.

I'll try to post the shots you requested later when I get some time.