[fredralphfred]

I've been reading reviews of the Tamron 17-50 vs. Sigma 17-50 & Pentax 16-50. One of the big complaints I read about the Tamron is that it has "trouble" focusing in low light. If they're all the same aperture (f/2.8), why would one lens focus "better" in low light than the others? Could this just be people's impression of the Tamron because of its relative cheapness, or is there something specific that would cause one lens to outperform another for focusing in low light?

Also, do these problems disappear when using contrast-detect (live view)?

I'd really love some sort of science-based explanation.

[Adam]

Bear in mind that AF systems weren't as advanced back then, so it's possible that this is no longer an issue. As for the cause, it could be something as simple as slightly less contrast wide-open compared to the other lenses.

[rawr]

Anecdotal reports of 'this lens is better than the other in low-light AF' should be considered with caution. Few of those making those reports use their lenses the same way - same camera, same lens, same scene.

What I mean is that in field use, what impacts the low-light focus performance of a lens might be a wide range of things:

- the camera AF settings (AFS, AFC, spot/all AF points, focus assist light on/off),
- the camera metering settings (spot/centre or matrix)'
- the contrast of the scene (generally degraded by low-light),
- the features of the scene (lots of small details, lots of varied light causing problems),
- the type of lighting in a scene (tungsten may still cause autofocus problems with early Pentaxes, for example),
- the T-stop of the lens (some lenses are naturally 'darker' than others),
- the use of 'protection' or UV filters on a lens, often dimming or murking it up,
- the flare and ghosting performance of a lens, which might cloud the lens sometimes'
- whether the lens has been adjusted properly for front or back focus.

etc etc. And often a combination of all of the above.

Hence I think it's hard to confine the discussion to just 'this lens is better than the other under low-light'. Best way to do it is under lab conditions, but hardly anyone does such specific AF testing, even the lens review sites like lenstip, Imaging Resource, Photozone don't.

BTW, I have the Tamron 28-75 and 17-50 f2.8's. Both seem to do OK in low-light.

[LensBeginner]

Also one man's low-light isn't necessarily the same low-light as the next man's...

[TheOneAndOnlyJH]

One thing I could think of might be slop or play in the focusing years. (I'm talking lenses in general, nothing specific to the lenses you mention.) If one lens has tighter gearing it might stop exactly where the camera tells it to, whereas another might stop just shy of where the camera feels it to. I know I've turned the screw drive of a lens using a precision screwdriver and you can wiggle it back and forth a couple degrees before the focus actually changes. I have no idea if there is as much play in lenses with built in SDM or HSM motors.

Another thing that might change people's impressions is sound. The Tamron will make a couple audible zip, zip noises as it settles in around the focus point. The Sigma does too but you can't hear it, which might make it seem faster to some people.

I have both the Sigma and Tamron and to be honest they seem to focus about the same in low light . In bright light I sometimes feel like the Sigma might maybe nail focus just slightly better, but only when pixel peeping (and it could also be due to my AF adjust settings). I don't think it's enough to justify the extra cost though.

The Sigma is also bigger and bulkier, and I can't stand that the zoom ring goes 'the wrong direction'. (Opposite all my other lenses.) Where the Sigma does excel is in being quiet. One of the reasons I haven't gotten rid of it yet is because I want a lens I can use without annoying other people, but even so I think I'll be letting it go soon because optically it's not any better than the Tamron.

[UncleVanya]

I will offer a bit of information that may or may not relate to the problem but seems relevant. I have had a couple of conversions where the SDM was not dead but focusing was unreliable. The lens would respond but not lock focus. Converting these from SDM to screw drive fixed the issue. Why? My assumption is that the SDM was not giving the level of precision that the focusing system needed and so it was not locking in perfectly. The screw drive provides more camera control and a different type of back and forth hunting even in lenses without problems focusing can behave slightly different on both systems. I wonder if the same is true of the low light use of these lenses - if the feedback given by the lens - the distance and precision traveled etc varies by brand. In any case that's all I can do to contribute as I don't own any of these other than the 16-50 screw drive converted.

[ivanvernon]

I've been reading reviews of the Tamron 17-50 vs. Sigma 17-50 & Pentax 16-50. One of the big complaints I read about the Tamron is that it has "trouble" focusing in low light. If they're all the same aperture (f/2.8), why would one lens focus "better" in low light than the others? Could this just be people's impression of the Tamron because of its relative cheapness, or is there something specific that would cause one lens to outperform another for focusing in low light?

Also, do these problems disappear when using contrast-detect (live view)?

I'd really love some sort of science-based explanation.

My Tamron 17-50 is terrible to focus in low light. I will not even take it with me any more to evening parties and events, just too frustrating.

[TheOneAndOnlyJH]

Also, do these problems disappear when using contrast-detect (live view)?

In bright light live view contrast detect should be immune to needing AF adjustment, but in very low light I've found that I get better results from PDAF. I'm assuming it's because the high ISO noise masks detail needed for accurate CDAF.

The Tamron 17-50 is parafocal, so the focus distance stays the same when you zoom (as opposed to varifocal where the focus distance changes). One trick is to zoom in to enlarge detail for easier focusing and then zoom out without refocusing. (Should work with PDAF and CDAF) I believe the Sigma is parafocal too, but don't quote me on that.

---------- Post added 10-23-17 at 08:35 PM ----------

My Tamron 17-50 is terrible to focus in low light. I will not even take it with me any more to evening parties and events, just too frustrating.

I mostly use Sel-2 9pt tracking on my K-3 using back button focus. I can still choose the focus target, and 9pt is a bit faster than 27pt auto. For my uses it works reasonably well in low light. (Full 27pt auto often seems overwhelmed in low light.)

[Jeff Lopez]

Other consideration would be the refractive index of each lens, of particular interest the reflectivity of the manufacturer glass used on that lens..
Wider lens were not made to have better focus, but to counter higher iso and slow shutter speeds. You might have a Sigma 2.8, which could be a dud for low light, and have a Zeiss 4.0, which would be a star.
Have a test with a smartphone with a fixed apeture. At day, it will focus right on spot, but at night it will pretty much hunt, until it decides what is in focus.

[EarlVonTapia]

Anecdotal reports of 'this lens is better than the other in low-light AF' should be considered with caution. Few of those making those reports use their lenses the same way - same camera, same lens, same scene.

What I mean is that in field use, what impacts the low-light focus performance of a lens might be a wide range of things:

- the camera AF settings (AFS, AFC, spot/all AF points, focus assist light on/off),
- the camera metering settings (spot/centre or matrix)'
- the contrast of the scene (generally degraded by low-light),
- the features of the scene (lots of small details, lots of varied light causing problems),
- the type of lighting in a scene (tungsten may still cause autofocus problems with early Pentaxes, for example),
- the T-stop of the lens (some lenses are naturally 'darker' than others),
- the use of 'protection' or UV filters on a lens, often dimming or murking it up,
- the flare and ghosting performance of a lens, which might cloud the lens sometimes'
- whether the lens has been adjusted properly for front or back focus.

etc etc. And often a combination of all of the above.

Good list.

Also add to this

- The AF module of the camera body.

Older camera bodies (like my K30) had the SAFOX IXi+ AF system which could operate at -1EV. Newer ones (like in my KS2) have newer modules, like the SAFOX X system that can operate at up to -3 EV. In my non-lab testing, I've found that indeed the KS2 focuses quicker and more accurately in lower light than the K30 using the same lens.

[rawr]

- The AF module of the camera body.

+1. I forgot what is probably the most important thing to consider.

If the camera AF system wasn't designed to work below a certain light level (in my case, K-x or K-5), the lens itself hardly matters.

FWIW, here's the low light EV ratings for the AF of a few recent Pentax cameras:

+2 EV: 645D
+1 EV: K20D, K200D, K-01
-1 EV: K-x, K-5, K-50, K-30, K-S1
-3 EV: K-5II/s, K-3, K-3 II, K-S2, K-70, K-P, K-1, 645Z

[leekil]

Why is the 645D worse than the K20D, when it came out after it? Does the +2 EV take into consideration that it would get less light due to the 645 lenses having a smaller maximum aperture?

[rawr]

Why is the 645D worse than the K20D, when it came out after it?

One reason may be that it had a different AF module from the K20D. The 645D shared the same SAFOX IX+ AF module as the K-5.

However it remains a bit of a puzzle to me why the 645D EV specifications are different from the K-5, even though both are SAFOX IX+.

SAFOX AF

11 AF POINTS:

K20D - SAFOX VIII
K200D - SAFOX VIII
K-7 - SAFOX VIII+
645D - SAFOX IX+
K-5 - SAFOX IX+
K-30 - SAFOX IXi+
K-50 - SAFOX IXi+
K-S1 - SAFOX IXi+
K-5II - SAFOX X
K-S2- SAFOX X
K-70 - SAFOX X

27 POINTS:

K-3 - SAFOX 11
645Z - SAFOX 11
K-3II - SAFOX 11
KP - SAFOX 11

33 POINTS:

K-1 - SAFOX 12

[EarlVonTapia]

However it remains a bit of a puzzle to me why the 645D EV specifications are different from the K-5, even though both are SAFOX IX+.

The 645D has a CCD sensor. The K20D (and the K5 and later DSLR models after it) has a CMOS sensor. That might have something to do with it.

[angerdan]

+1. I forgot what is probably the most important thing to consider.

If the camera AF system wasn't designed to work below a certain light level (in my case, K-x or K-5), the lens itself hardly matters.

FWIW, here's the low light EV ratings for the AF of a few recent Pentax cameras:

+2 EV: 645D
+1 EV: K20D, K200D, K-01
-1 EV: K-x, K-5, K-50, K-30, K-S1
-3 EV: K-5II/s, K-S2, K-70, K-P, K-1, 645Z

Great list!
Where do you get the data? Pentax/Ricoh seems to state different values for low light EV AF working range.