[Smeggypants]

I had a couple of friends who had the Sigma 30mm (in Canon, Nikon and Pentax mounts) and they all encountered varying degrees of focus issues at close distances. They had to send their lenses to be recalibrated at the local Sigma service center. It seems to be pretty common where I am because Sigma charges a fee of S$80 for it. Friends who got the early batch of the Sigma 50mm in Canon and Nikon mount also faced this issue, which is easily adjustable by Sigma. Perhaps you might need to check with your local Sigma service center if you're not getting precise focus.

It's cheaper to set it to '+4' on the camera's AF calibration than pay S$80 to have it sitting at '0' though.

[creampuff]

It's cheaper to set it to '+4' on the camera's AF calibration than pay S$80 to have it sitting at '0' though.

That's definitely true. Sigma's QC is known to be erratic and the FF issue with some copies of the 30mm is well known. In the case of the 50mm, the issue was very obvious erratic focus in the early batches at distances of 3m and beyond.

[betaPhoto]

Enclosed is a set of photos with phase AF, DFA 100mm 1:2.8 at f=2.8, subject distance 1 m, direct light from a 'naked' 25W tungsten bulb where distance was adjusted to vary light level as seen by the camera using spot metering, most of the light should be directly from the bulb, some has inevitably been reflected from the surfaces of the room, though. It was dark outside so all light in the otherwise dark room was from the single bulb that was mounted at on a tripod at the approximate heigth of middle of the target. This was kept on the right side on the camera at about 30-40 degree angle to the line from the sensor to target.

#1 EV 4 (t = 0.5s) : camera WB 3308K : focus OK
#2 EV 3 (t = 1.0s) : camera WB 3278K : front focus
#3 EV 2 (t = 2.0s) : camera WB 3148K : front focus (somewhat worse than #2)

With the use of spot metering the area relevant to AF operation should be within 1/3 EV of the above values assuming camera meter is accurate.

How is the focus at anywhere between EV 8-14 with the same setup. In other words, is the DFA 100mm on the K-5 focus set correctly for outdoor and bright indoor light levels?

[jolepp]

Two more test shots with phase AF, DFA 100mm 1:2.8, f=2.8, center point focus etc. as above, light sources indirect daylight and indirect light from a 7 W CFL (nominally 3000K) instead of the tungsten bulb, for the eye the light seems much whiter than the tungsten bulb.

#1 indirect daylight : EV 0.65 (t=5s) : camera WB 7525K : focus OK
#2 CFL : EV 0.39 (t=6s) : camera WB 3339K : focus OK

So it seems that with the right kind of light my K-5's phase autofocus can work just fine, even in low light. Direct light from the CFL results in much higher EV values with which my K-5's phase AF seems to work just as well as at EV 0.39. Unfortunately tungsten light is a problem.

YMMV.

---

Last edited by jolepp; 03-03-2017 at 02:40 AM.

[jolepp]

How is the focus at anywhere between EV 8-14 with the same setup. In other words, is the DFA 100mm on the K-5 focus set correctly for outdoor and bright indoor light levels?

I haven't really used the lens in proper daylight; I got it fairly recently and there isn't really much to point it to outside this time of year. I ended up setting AF fine adjust to +2 in daytime inside light through a window (this would have amounted to maybe EV 6). I cannot get very much higher than 4EV with that naked 25 W bulb. I have tried a 40W tungsten reflector one for brighter tungsten light and from that my impression is that AF works ok in bright tungsten light; I got the 25W bulb and a holder for dim enough direct light to be a problem, actually . (The idea being creating something resembling repeatable conditions like those where I have had actual problems).

[Smeggypants]

#1 EV 4 (t = 0.5s) : camera WB 3308K : focus OK
#2 EV 3 (t = 1.0s) : camera WB 3278K : front focus
#3 EV 2 (t = 2.0s) : camera WB 3148K : front focus (somewhat worse than #2)

Exactly my findings!!

[Tommot1965]

I've done my test under Halogen lights, using a dimmer switch to alter the EV for subsequent shots ...I see FF at EV 4 under these conditions...its minor, but there none the less

thanks to Jolepp for supplying the test sheet, rep added

shot parameters were , shake reduction off, tripod mounted with remote release, ISO 100......subject was 500mm from the Focal plane , WB AUto

at EV 2 the af lamp assist came on, so I disabled it

I did a sequence of three shots

1 View Finder Phase detect
2 View finder phase detect with Mirror lock up
3 Live view

the MLU shot showed no difference, so I will not show those in the post..

EV of the tests were 2,3,4,5

lastly I tested EV4 { were for me FF was present at F2.8} at F4....

the dof was razor thin....10mm at F4 it went to 14mm

the second shot in each sequence if LV for reference

EV2





EV3





EV4





EV5





EV4 @ F4

---

Last edited by Tommot1965; 01-29-2011 at 05:51 PM.

[Ray Pulley]

I've done my test under Halogen lights, using a dimmer switch to alter the EV for subsequent shots ...I see FF at EV 4 under these conditions...its minor, but there none the less

thanks to Jolepp for supplying the test sheet, rep added

shot parameters were , shake reduction off, tripod mounted with remote release, ISO 100......subject was 500mm from the Focal plane , WB AUto

at EV 2 the af lamp assist came on, so I disabled it

I did a sequence of three shots

1 View Finder Phase detect
2 View finder phase detect with Mirror lock up
3 Live view

the MLU shot showed no difference, so I will not show those in the post..

EV of the tests were 2,3,4,5

lastly I tested EV4 { were for me FF was present at F2.8} at F4....

the dof was razor thin....10mm at F4 it went to 14mm

the second shot in each sequence if LV for reference

Tommo,

The EV4 and below shift is pretty consistent it seems. BTW, the K5 specification for AF is -1EV to 18EV. Mine would not consistently turn on the assist until well below EV0 using the PDAF. The LVAF turned on the light well before that.

What focal length lens did you use?

Thanks,

Ray

[betaPhoto]

...
at EV 2 the af lamp assist came on, so I disabled it ...

Do you think the new firmware (1.02.21.10) is having the af assist lamp come on sooner? Or possibly some are slightly underexposed and EV 2 is actually closer to EV 1?

... BTW, the K5 specification for AF is -1EV to 18EV. Mine would not consistently turn on the asisst until well below EV0 using the PDAF. The LVAF turned on the light well before that.

What firmware are you using? 1.02.21.07 or 1.02.21.10.

[Tommot1965]

Firmware is the last 1.02 silent release , lens was sigma 17 -50 @ 50 mm

As for the EV,

The EV was set using the VF matrix metering, and it was bang on, although when using Lv the metering would alter a tad a 2 second shutter using VF would drop to 1.6 secs using Lv under the same light

I'm not sure if the new firmware switches the AF lamp on sooner. I've never tested that, but it doesn't seem to

[Tommot1965]

ok

after reading reports of better performing K5's in low light, and after talking to my distributor, I exchanged my 27/11/10 built K5 for a 17/1/11 K5 ....roughly 3 months between them..

and my new K5 does perform better in the same situation as the previous one !!!.

so there are definitely better performing K5's out there.. whilst still not prefect....it does get perfect by F4...or greater distance to the subject..thus increasing DOF

shots as as before, Halogen Light source. F2.8 @ 50 mm ISO 100

I didnt include the LV images...these are all VF phase detect.

I have included two examples from the previous K5 in this post for a comparison ..@ EV3 and EV2....

Im now done testing...:}

EV3



Previous K5 @ EV3




EV2




Previous K5 @ EV2



EV1.5



EV0

[RBellavance]

Enclosed is yet another focus test sheet. The intent is that this should be a simple and cheap aid for detecting/demonstrating front/back focus: this is basically a A4 sized document that has dotted lines to ease folding the printed sheet (with a ruler) to resemble a "stair" standing on its side where the width of the difference in depth of middle/side sections can be varied in 1/2" steps.

Nice and simple, thanks !

[jolepp]

Below is a series of test shots to demonstrate the effect different light sources have on K-5 AF. The conclusion is that tungsten light as such might be the problem as opposed to light of similar color temperature in general: the snaps below show FF in tungsten light at EV 3 / 2 while the ones taken in the same conditions except that the 40W tungsten bulb was replaced with a CFL / a LED based bulb are OK. My current hunch would be that this is related to the continous, IR-rich spectrum (*) of tungsten light; AFAIK the alternate bulbs produce much less IR (this is why/how they use much less electricity to produce the same amount of light in the visible spectrum).

I used a generic 40W E14 tungsten bulb (**), for alternate bulbs I used Philips offerings with the idea that these might actually be available elsewhere in case this turns out to have relevance (as in someone else being able to repeat my experiment). The CFL is a 8W Philips 'Turbo Energy Saver' ('TORNADO WW827 E14 8W 8000h 500 lumen' EAN 8 710163 214283) and the LED item is a 2W Philips LED 'MyAccent' (20.000h, 86lm, warm white, EAN 8 727900 844856).

Below, the stripe of garbage at the right edge of the frame is an articact from ufraw 0.17; it sometimes does this with the K-5 DNGs. Extracting the embedded .JPG from the DNGs and resizing that for posting would have gotten rid of this, but it also got rid of the EXIF info, so I tought I'd rather have the stripe than risk goofing up with extracting and attaching the EXIF .

Edit: the test target has been folded for a 1" difference in depth between the middle and side sections.

(*) I owe this insight to this post: https://www.pentaxforums.com/forums/#post1370648

(**) As far as repeatability goes I would not worry too much about using just about any tungsten bulb, at any rate CFLs and LEDs are far more heterogenous.

[jolepp]

Direct light from a 40W tungsten bulb at right to camera, ~45deg angle from lens axis, distance varied to adjust intensity, DFA 100mm 1:2.8, f=2.8, target distance to sensor plane 1m, assuming camera metering (spot) correct the EVs given should be accurate within 1/3 EV

#1 : EV 4 (t=0.5s) : focus OK : camera wb 3278K
#2 : EV 3 (t=1.0s) : FF : camera wb 3251K
#3 : EV 2 (t=2.0s) : FF : camera wb 3278K

[jolepp]

Direct light from 8W Philips CFL ('Turbo Energy Saver' ) at right to camera, ~45deg angle from lens axis, distance varied to adjust intensity, DFA 100mm 1:2.8, f=2.8, target distance to sensor plane 1m, assuming camera metering (spot) correct the EVs given should be accurate within 1/3 EV

#1 : EV 4 (t=0.5s) : focus OK : camera wb 3100K
#2 : EV 3 (t=1.0s) : focus OK : camera wb 3114K
#3 : EV 2 (t=2.0s) : focus OK : camera wb 3161K