[jeffshaddix]

Hey guys,

I've got a brand new K-5 with a significant front focus problem under low light and/or tungsten light. The problem seems to dissipate as the scene lighting increases. The problem is only exhibitted in PDAF, and can be seen with both AF lenses and MF lenses with the focus confirmation; contrast AF and manual focusing w/ focusing screen is fine. AF microadjust would need some -40 setting (which isn't available) to fix the FF in low light, but then of course causes back focus in daylight.

Clearly there's a problem with either the software or hardware related to my light wavelength sensor and PD AF module. At first I thought, "Hey no problem, I'm only using FW 1.01 and this was fixed with FW 1.03." So I updated to 1.13, but to no avail, still broken.

Thoughts? Should I send it to CRIS, or is this normal behavior? Falk's report makes it sound normal, but other users say otherwise and that it's repairable.

Has anybody sent their camera to CRIS for this? Did it help? How long did the repair take?

Thanks!

[joe.penn]

At first I thought, "Hey no problem, I'm only using FW 1.01 and this was fixed with FW 1.03." So I updated to 1.13, but to no avail, still broken.

You just bought it and it was loaded with FW1.01? That is, well, strange (I would think any new units would at least have 03 or better on it)...

[Raffwal]

Should I send it to CRIS, or is this normal behavior? Falk's report makes it sound normal, but other users say otherwise and that it's repairable.

After experimenting with my K-5 silver and reading many many threads on the subject, I'd say it's normal behavior for Pentax. Sorry.

[morpho]

What is tungsten lighting? Like stagelights?
I shot an indoor festival this weekend, about 1000 shots fired. Granted, for most of the time, the lighting conditions were very good for a concert, but i can't say any of my shots have suffered from any misfocusing that was the cameras fault. I've heard of this being a problem before, but i can't say i've encountered it.

I was the only one with a Pentax out of around 20 photographers. I talked a lot to one guy who had a D3s and a D800. At one point when we were waiting to be allowed into the photo pit, i tried taking a picture of the guard, but there was too little light; the AF wouldn't lock. I asked the guy with the D3s and D800 if he could lock focus on the same subject. He couldn't do it either.

I don't know if this has any relevance to what you are experiencing, but can it be that you're expecting the camera to do something it's not supposed to do?

I also found that for the best results when shooting moving artists and getting the focus allright on the eyes, i aimed the square a little to the side of the eye first and movid it inwards towards the eye in AF-C. For the most of the time, it worked. Perhaps the large central focus point is messing a bit with your mind?

[Deimos]

FYI my K-5 needed no lens AF adjustment with 1.01 (which is what it came with brand new direct from Pentax in Nov 2011) but I need -3 with 1.13 for my FA43 so you may also need some AF adjust with the new firmware

I have noticed some FF in very very low light or strong tungsten, much better than the k-r was though. If it happens I just use LV

[jbinpg]

Before you ship it off, try resetting the camera and leaving all AF adjustments at zero then do some tests. Many people have reported much better AF by doing this with firmware 1.13.

[jeffshaddix]

Joe - I agree, weird that the factory is still using 1.01.

Raffwal - hrm, OK so when does your K-5 start to struggle? Mine is inaccurate at night indoors with a 60-100 watt incandescent lighting a room.

Morpho - tungsten is the filament used in incandescent lightbulbs, it produces a reddish light that's often hard on cameras to process (for white balance and PDAF). I thought it might be a problem with the center focus point size, but I switched to a test chart designed for it and it's the same behavior. The reason I expect accurate behavior is that it worked fine for my K-7.

Deimos - yeah I can always just manual focus or use LV, but if this is fixable why not ya know? Thanks!

Binpg - Well you got my hopes up. I tried doing reset from the settings menu, reset custom functions, and reset within the AF adjust menu all to no avail. Is there another system-wide reset I'm missing?

Thanks for the replies guys - it's frustrating that this issue is so hard to pin down.

[rawr]

Assuming that the issue may be that the tungsten light spectrum has some IR in it that the K-5 AF sensor has difficulties dealing with... is this one of those occasions where a UV filter may actually be useful, due to it's ability to filter out (to some degree) IR spectrum light?

Pic related, from the lenstip.com review of UV filters.

[jeffshaddix]

Good thought rawr. Unfortunately the UV filter on/off doesn't look like it makes a difference. Doh!

I'm curious if anyone has ever sent their camera in for repair for this issue (and the outcome)?

[crewl1]

Does the problem occur with white balance setting in auto as well as tungsten? I had heard somewhere that if you set the WB to Tungsten it alleviates the issue.

[Ray Pulley]

Hey guys,

I've got a brand new K-5 with a significant front focus problem under low light and/or tungsten light. The problem seems to dissipate as the scene lighting increases. The problem is only exhibitted in PDAF, and can be seen with both AF lenses and MF lenses with the focus confirmation; contrast AF and manual focusing w/ focusing screen is fine. AF microadjust would need some -40 setting (which isn't available) to fix the FF in low light, but then of course causes back focus in daylight.

Clearly there's a problem with either the software or hardware related to my light wavelength sensor and PD AF module. At first I thought, "Hey no problem, I'm only using FW 1.01 and this was fixed with FW 1.03." So I updated to 1.13, but to no avail, still broken.

Thoughts? Should I send it to CRIS, or is this normal behavior? Falk's report makes it sound normal, but other users say otherwise and that it's repairable.

Has anybody sent their camera to CRIS for this? Did it help? How long did the repair take?

Thanks!

My K-5 FF under low tungsten light was corrected with firmware 1.03. I have not updated it past that point as I do not need any of the stuff that was in later firmware and I am happy with the AF and want to stay that way.

I have extensive tests with a lens align before/after 1.03 that clearly show the rearward offset after the firmware was installed.

BTW, I do not agree with Falk on this one based upon my tests and many real world shots before and after 1.03 (as well as extensive tests and shots with 3 K5s before 1.03). My K-5 does miss focus just a bit occasionally, and it seems to do so on a red or pinkish subject like a face filling the frame. This is pretty rare, however.

That being said, you mentioned 60-100 watt tungsten lighting conditions. Depending upon what lens you have on the camera, you could be outside of the specified range of the AF system.

The range low limit is -1 EV ISO 100 with the 50mm f1.4.

If you are using a slower lens than the 50mm f1.4, the low limit moves up accordingly because the AF module only gets as much light as the maximum aperture of the lens allows into the mirror box. For example a lens with a maximum aperture of f4.0 would move the lower limit to +1 EV and so on.

I have another body (new) on order and it will be interesting to see how it performs compared to my current body.

I suspect that you will need to send your camera in, but it would be useful to note the actual exposure data along with the lens used.

As for IR:

This theory has been kicking around since the original Pentax DSLRs were introduced as every single one of them (I have owned many of them) moved to FF in low tungsten light.

Hot mirror IR filters (expensive) have been tried with no improvement going back as far as the K10. I believe that the AF system already has an IR filter built-in anyway. The issue has never been fully explained for any of the bodies, at least as far as I am concerned. The K-5 just happens to have a color sensor built-in to the AF system, so Pentax could adjust the firmware to try and offset the AF system for the problem (which does seem to be related to color, and in this case reds), which I gather was the reason for this sensor in the first place.

It is interesting to note that the Nikon D7000, released about the same time as the K5, also has some sort of color sensing element in the AF system, and there are many complaints of focusing errors with it as well, but BF instead of FF. Apparently, whatever optical problem they were trying to solve is part and parcel to PDAF systems.

Ray

Red backgrounds in low light are tough. This was done with a Sigma 100-300/4 at f4, ISO 3200 and 1/320 about 70 feet from stage. He was doing a series of spins. It would've been tough with my 50-135 too. I don't know why but after shooting tens of thousands of exposures in rehearsals and performances, when they light it red, its tough. My hunch is its contrast. I manage. The video guy has pro Sony stuff and plenty of color noise issues and he's at f1.8. Hot lights in theater are mostly Tungsten, starting to see some led stuff. When I can get closer during dress rehearsals, its better. Back of the house, red is tough.

[jeffshaddix]

Crewl - Just tried tungsten WB and it seems to have no effect. Good thought though...it certainly made the white balance better than with auto.

Ray - I looked at the exposure information this morning. At ISO 100, I'm shooting 1/15 s @ f/1.4. I think that corresponds to a scene Ev of 5, but of course that's tungsten light, which Falk thinks is harder to detect by the PDAF sensor.

Brooke - Nice shot! For situations like that I may be inclined to just use manual focus.

Crewl - Just tried tungsten WB and it seems to have no effect. Good thought though...it certainly made the white balance better than with auto.

Ray - I looked at the exposure information this morning. At ISO 100, I'm shooting 1/15 s @ f/1.4. I think that corresponds to a scene Ev of 5, but of course that's tungsten light, which Falk thinks is harder to detect by the PDAF sensor.

Brooke - Nice shot! For situations like that I may be inclined to just use manual focus.

You might be inclined but you'll miss it. Constantly zooming in and out to fill frame with dancer without lopping off limbs. They move a lot and often. I knew this was coming, locked on and followed, caught just her as she is caught. Lots more examples here Cary Ballet Company | Brooke Meyer

[Ray Pulley]

Crewl - Just tried tungsten WB and it seems to have no effect. Good thought though...it certainly made the white balance better than with auto.

Ray - I looked at the exposure information this morning. At ISO 100, I'm shooting 1/15 s @ f/1.4. I think that corresponds to a scene Ev of 5, but of course that's tungsten light, which Falk thinks is harder to detect by the PDAF sensor.

Brooke - Nice shot! For situations like that I may be inclined to just use manual focus.

That should be about EV5.

I understand the reasoning in Falks study but I do not agree with his conclusions or results based upon my own tests and real life shots in light where the problem shows up.

Regardless, take a look at the quick test I just posted on my second K5 body. The results are 100% in accordance with my first body, and in both cases the firmware update obviously moved the focus point backwards in low tungsten light. In both cases my bodies are focusing accurately in light where they would obviously FF before the firmware fix.

Send your K5 in for repair or exchange it if you can. No camera is perfect and my K5 will mis-focus occasionally (reds as I noted earlier), but after the firmware fix for FF, my K5s are far better than all of my earlier Pentax DSLRs (2 each: istds, k100, K10, K20, 1 each: K100 Super, K7), all of which had FF in lower tungsten light and all of which would not focus at all in light where the K-5 gets a lock.

Ray