[Nightwings]

I've been going nuts for the past week trying to figure out if there is something wrong with the K-x autofocus system. I've always thought my 18-55 kit was a little too soft at the 55mm end wide open. So I've been doing all kinds of focus tests with other lens's I have. In particular my SMC F 50mm 1.7 prime.

Here's what I don't get.

I set up a 6 x 6 inch text target 3 feet away from the camera ... tripod mounted, remote activated. F1.7 .. wide open. The autofocus consistantly had it front focusing.... making the text blurred. I went into debug mode and found the perfect correction of -150. Autofocusing was bang on razor sharp. So then, I take the camera outside and start taking some shots. All shots taken farther than 10 feet or so were blurred. I'd have to stop down to f11 or higher to get resonable results.

Here's what I can't figure out. If my fine tuning made an obvious clear improvement on the target, why the h*ll didn't it hold the correction throughout the range of the lens?? Also ... even before the correction, I would go into live view and focus from there using contrast detection and it focused beautifully. Surely this is a defect that Pentax must know about? Why would they use phase matching when the alternate contrast detection clearly works better?

Am I chasing an imaginary boogyman? Am I being to fussy?

[dgaies]

One possibility is that you adjusted your focus under tungsten lighting (indoors) and then took shots outdoors. Most Pentax DSLR bodies (less so with the K7) will front focus to some degree under tungsten light. So you might have added a focus correction to a lens that didn't need it. You could confirm this by shooting a far away target indoors under the same type of light you made you initial correction with. Or, you could redo your focus test outdoors, adjust the focus correction under natural light, and then try shooting a far away target.

[rawr]

I've not experienced any focus problems with the kit 18-55.

Pls remember that any testing procedure has to be performed very carefully and clinically. This forum has literally dozens and dozens of discussions on the topic of doing AF tests, and just as importantly, doing AF tests wrong.

And any exposure adjustment of -150 is pretty extreme. Go easy there. That's about a country mile of adjustment. Small wonder the lens goes wonky once you actually start to use it.

[Nightwings]

Most Pentax DSLR bodies (less so with the K7) will front focus to some degree under tungsten light.

GOOD GRIEF! You can't be serious??!! Of course you are. Wow, that is a MAJOR design flaw if you ask me. Here I am slamming my head against the wall trying to figure this out - Why on earth would it ever occur to me that .... in 2010 ... I'm dealing with a camera that hasn't figured out how to focus properly under ..... the first ever form of indoor lighting invented. Do Nikon's, Canons etc. have this quirk also? I've always owned Pentax camera's since my first spotmatic ....... but THIS? ...... I'm speechless.

[dragonfly]

I don't think candles were ever made out of tungsten.

That was just a joke.

As it is, the kit lens leaves a lot to be desired. Your F lens should be sharp at the camera's standard focusing setting.

Also, where is Winterland? Is that in SF?

[rawr]

Users of every single camera brand have reported AF issues, real or perceived.

You may find it useful to have a look at what Canon say about AF testing and some of their suggestions re AF adjustment for some of their pro bodies:

http://www.usa.canon.com/uploadedimages/FCK/Image/2007/MARK%20III%20Suggeste...ings_Final.pdf

They say lots of useful stuff, like if you are a landscape shooter, test your AF outdoors against a typical outdoor scene, not by (for example) shooting a ruler indoors.

[Marc Sabatella]

GOOD GRIEF! You can't be serious??!! Of course you are. Wow, that is a MAJOR design flaw if you ask me.

Perhaps, but it's is flaw of pretty much all phase detect AF systems, not just Pentax.

But the discrepancy is pretty small. Shouldn't really ever be noticeable with the kit lens, as the DOF isn't small enough to show a discrepancy that size. I suspect you're seeing something else.

[dgaies]

I set up a 6 x 6 inch text target 3 feet away from the camera ... tripod mounted, remote activated. F1.7 .. wide open. The autofocus consistantly had it front focusing.... making the text blurred. I went into debug mode and found the perfect correction of -150. Autofocusing was bang on razor sharp. So then, I take the camera outside and start taking some shots. All shots taken farther than 10 feet or so were blurred. I'd have to stop down to f11 or higher to get resonable results.

Shouldn't really ever be noticeable with the kit lens, as the DOF isn't small enough to show a discrepancy that size.

Based on the OP, I think he was testing and adjusting the focus at f/1.7 (using an F50/1.7).

[Marc Sabatella]

You're right, my mistake - I didn't read nearly carefully enough. In which case, then yes, that's probably exactly what he's seeing.

Also, to answer one of the other questions in the OP: all DSLR's use phase detect because a) it's a ton faster than contrast detect, and b) contrast detect can only work more accurately when the actual imaging sensor is being used to form the image, which means it can only work in liveview mode

[Wheatfield]

When I got my K20, I dutifully adjusted all my lenses to the Nth degree with a focus chart from 1-2 meters away.
And all my pictures were buggered because the focus wasn't right.
So, I came up with a different method.
I now adjust my lenses with a target that is 3-5 meters away, and lo and behold, everything is peaches again.
Moral: If all you do is macro, adjust your lenses at macro distance. If you tend to do more general photography, then adjust your lenses at a median focus distance.
And, most Pentax AF sensors are not colour blind.
The K7 sensor is pretty colour blind.
I don't know how the sensors from other makers do in this regard.

[Wesbo]

If you make focus adjustments indoors under tungsten lighting at a close target, it will not be optimized for a target farther away under daylight. This is true to some degree for every phase-detect autofocus system. End of story.

Contrast-detect is superior under some circumstances, as you observed, but it is slow and AFAIK is ignorant of two things that make it inferior under some circumstances:

1. How far out of focus a subject is (that is, how much focus correction is needed)
2. Which direction correct focus lies (that is, which direction, near or far, to focus)

I suspect you could re-do your focus corrections outside using a target farther away, and would see good results.

[Nightwings]

Thanks for the tips guys .... I guess I will have to use a target further away. ... And depending on the type of target .... seems to affect the results. But yes, I will be doing further tests outside. The K-x is my first digital DSLR and now that I can pixel peep, I guess it gets me paranoid .. the type of paranoia I never experienced with my film SLR's. On the other hand ... My FUJI S100fs is smack on ... razor sharp focusing one end of it's 14x zoom to the other. As much as I like the K-x .... they'll have to pry my S100fs from my cold dead hands

[Nightwings]

Also, where is Winterland? Is that in SF?

Canada

[notsowise]

What is debug mode?

[rawr]

...The K-x is my first digital DSLR and now that I can pixel peep, I guess it gets me paranoid .. On the other hand ... My FUJI S100fs is smack on ... razor sharp focusing one end of it's 14x zoom to the other.

I don't understand any paranoia. It's just a camera, not some sort of magic box. Relax a little.

And you do understand that the end-to-end sharpness you are seeing with your Fuji may have as much to do with the vastly deeper depth of field you get out of a smaller point and shoot's sensor than any wonderfulness of the Fuji AF.

As the owner of two Fuji point and shoots - an F10 and an F200 - I can say with confidence that the Fuji AF isn't infallible or anything especially amazing.

I also presume that you have been conducting your K-x AF tests with spot focus (ie using the centre AF point only), not any other mode. You need to do it that way for your results to be reliable and consistent.