[Murfy]

Has anyone used this method of fixing front and back focus? And if so, did it work?

A Thousand Words A Picture: How to Use DotTune AF Fine Tune Calibration For Nikon Cameras

[vonBaloney]

It isn't the only way, but yeah it works. One thing I found out doing this is that on my K-5 (and so probably all others), the aperture setting will affect how readily the green hexagon will confirm focus, so the camera actually allows some extra leeway for depth of field when you stop down. (Even though with this method, you never actually need take a shot -- just setting the aperture on the dial.) In other words, if you are using an f/2.8 lens and leave the aperture dial at f/2.8, it will be much more picky about confirming focus than if you set it at f/8 where it will quickly confirm anything in the ballpark. Consequently you need to make sure you set the aperture to the max widest that your lens allows for greatest accuracy.

[stevebrot]

It isn't the only way, but yeah it works. One thing I found out doing this is that on my K-5 (and so probably all others), the aperture setting will affect how readily the green hexagon will confirm focus, so the camera actually allows some extra leeway for depth of field when you stop down. (Even though with this method, you never actually need take a shot -- just setting the aperture on the dial.) In other words, if you are using an f/2.8 lens and leave the aperture dial at f/2.8, it will be much more picky about confirming focus than if you set it at f/8 where it will quickly confirm anything in the ballpark. Consequently you need to make sure you set the aperture to the max widest that your lens allows for greatest accuracy.

What he said...

It also helps to make several estimates before settling. Even better, do several focus checks before deciding that the AF needs adjustment. If you have less than 6 of 10 front/back focused, adjustment will be an exercise in frustration. Test in strong light and use a flat (not slanted) high contrast focus target at 15x-20x the focal length distance from the target.


Steve

[vonBaloney]

What is interesting about the dot tune method is that it will tell you if you need adjustment or not because you're finding the center point, and interestingly enough without ever actually doing any autofocus as part of the testing procedure. So it can be worth it just for peace of mind knowing that your lens is calibrated so every time you do miss focus on something you're not wondering. On a zoom it is worth doing at several points. You do need to use an flat, isolated target as pointed out above so there is no possibility of confirmation on "target adjacent". Once you get it down, you can do the whole process in a couple of minutes...

[Murfy]

So right now, I have it at +7 where it is not in focus (+6 has a green dot confirmation, +7 does not) but green dot confirmation shows up on the other end at -10, I can't go any further down. This is for apply for one setting, at 50mm, on the 18-135 at f4/5. I was about 8 feet or so from a chair I was focusing on. I will pick this back up again on Monday and try some more.

I have no focusing issue in Live View, but I do with using viewfinder. I will also test with the 55-300 lens, I haven't tested that lens yet.

[vonBaloney]

So right now, I have it at +7 where it is not in focus (+6 has a green dot confirmation, +7 does not) but green dot confirmation shows up on the other end at -10, I can't go any further down.

Are you saying that the dot is on at +6 and -10 but not in-between those? That doesn't make sense. Also a chair is probably not the best target unless you are focusing square to a flat surface of upholstery maybe (with a high-contrast pattern). There should be a whole range of values (of maybe 4-8 points, but a slower lens will confirm more points) where the green dot will show (and those should be adjacent to each other), and some where it confirms instantly and some where it takes a second (which should be regarded as secondary confirmations) -- you're looking for the center point where an in-focus target is instantly and solidly confirmed (no flickering). Remember, set the aperture to max and make sure the system is off (not confirming anything) each time before you hit the shutter half-press.

Or, if you mean it is confirming between +6 and -10 (which should mean it is instant and solid around -2/-3), then I would put it there (-2 or -3) and see how it goes...

[mee]

One can go crazy micro-focus adjusting their lenses if they don't have the setup correct. And this dot focus, from reading the info on the provided link, doesn't seem to work as well for Pentax cameras as it would on the Nikon system apparently.. since they have arrows to tell if the camera thinks the focused object is in front or behind. We'd have to eyeball it since we don't have those arrows.

At which point I think I'd rather do the ruler at 45 degrees test.. but one must ensure they keep the camera at the proper height and distance away (there is a site that has the calcuator and steps.. but it alludes me at the moment) otherwise they risk faulty results. I've experienced this first hand where the lens is apparently back focusing, I adjust with the ruler to center. Test in the 'wild' it appears to be front focusing. Recalibrate with the ruler now it's off again.. sometimes front, sometimes back.. wait what?

In live view it will focus fine because its using a different focusing type... I typically use that as my baseline though with the timer set to minimize camera shake.


TL;DR: Ensure your system is setup properly to perform mf calibration or you risk going nuts.

[Murfy]

Are you saying that the dot is on at +6 and -10 but not in-between those? That doesn't make sense. Also a chair is probably not the best target unless you are focusing square to a flat surface of upholstery maybe (with a high-contrast pattern). There should be a whole range of values (of maybe 4-8 points, but a slower lens will confirm more points) where the green dot will show (and those should be adjacent to each other), and some where it confirms instantly and some where it takes a second (which should be regarded as secondary confirmations) -- you're looking for the center point where an in-focus target is instantly and solidly confirmed (no flickering). Remember, set the aperture to max and make sure the system is off (not confirming anything) each time before you hit the shutter half-press.

Or, if you mean it is confirming between +6 and -10 (which should mean it is instant and solid around -2/-3), then I would put it there (-2 or -3) and see how it goes...

the green dot confirms at -10, all the way to +6. I can't go past -10 to know if it would not appear at -11, because I can only go to -10 (I feel like there is a Spinal Tap joke in there somewhere) mine confirmed instantly on those. so at 0 to +6 it was solid green right away, at +7, there was nothing, no dot at all. from 0 to -10 there was also a constant green dot, with no hesitation or flickering. I waited for the meter went off before confirming again.

the chair has a thin high contrast flower pattern pad on it, but not flat as a wall, so I will test again with an actual test pattern, this was just a test for the process and I am very impressed how easy, in general, it was to do.

[derekkite]

I found the best way to get your lens AF calibration right is to send both in to a service center. Both come back, in my experience cleaned and calibrated.

I check by focusing on a contrasty item with some forward and back indication such as a 45 degrees ruler in live view. Then switch to pdaf focus. Then back to see if it is still in focus on live view. Focus peaking indicates where in the depth of field the camera focused on.

[stevebrot]

the chair has a thin high contrast flower pattern pad on it

Chair? Follow the instructions on the DotTune video page and download a proper focus target (link provided below). Tape the target to a wall and position the camera on tripod at 50x the lens focal length from the target. It is important that the sensor be parallel to the target.

You Tube

You Tube

Downloadable chart

http://www.testcams.com/DotTune/CrossHatch_Target.png

Print this chart on high gloss photo paper.



Steve

[Murfy]

Chair? Follow the instructions on the DotTune video page and download a proper focus target (link provided below). Tape the target to a wall and position the camera on tripod at 50x the lens focal length from the target. It is important that the sensor be parallel to the target.

DotTune: Autofocus fine tuning in under 5 minutes - YouTube

Downloadable chart

http://www.testcams.com/DotTune/CrossHatch_Target.png



Steve

yes I have one I printed earlier, but couldn't find it and I was impatient and curious to try the technique! I'll try again tommorow, properly. The process was much easier than others.

[Murfy]

I believe I got it to work! I had to adjust overall and then for each lens, but it looks like I have focus. If I notice it does not work, then off to Pentax for repair because I am sick of dealing with it.

[mee]

I watched the video and think this is interesting.. but I'm having poor results. It seems to get in the ballpark VERY quickly but Critical Focus in LV still appears sharper than adjusted normal shooting here.

I've tested this with a DA* 50-135mm at 50mm f/2.8 LV seems sharper sitting on a tripod targeting a static option (exterior molding in my situation, also a book indoors) at the +2 setting I got from doing the dot focus. Interestingly, 0 setting is just as sharp, if not slightly sharper than LV's test shots. Strange. I suspect SDM has a play in this.. but I cannot confirm that. Again, one could go crazy trying to fine tune their lenses. haha...

Now I'm going to try testing a couple of prime lenses... yep day is booked solid on this now.. ugh.

[stevebrot]

Again, one could go crazy trying to fine tune their lenses. haha...

I agree. DotTune is one of the better calibration methods out there, but anyone expecting PDAF to consistently match magnified LV is going to be disappointed. The precision of the PDAF system simply is not there, particularly for subjects with low contrast and/or complicated contours.

A few notes regarding PDAF focus tuning:

• Just because you see evidence of front/back focusing, don't jump to the conclusion that AF can be improved through in-camera tuning
• Adjustment based on a curved or slant focus target is possible, but ill-advised
• A high contrast focus target is essential
• Never tune based on hand-held tests
• Do not base your adjustment on a single measurement. DotTune is intrinsically good in that regard.
• When doing DotTune (or similar), a wide spread indicates a target with inadequate contrast, a lens with inadequate contrast/resolution, and/or inadequate light. If you have accounted for light and target, decrease the distance used for the test
• Even when properly "dialed in" using the focus fine adjust, a lens may still front/back focus at a different focal length (zooms) or subject distance (internal focus designs).
• Accurate fine tuning at 2 meters does not automatically translate to accuracy or precision at 20 meters. A good rule of thumb is to not expect the AF system to do a better job than manual focus using the optical viewfinder.

Steve

[bxf]

One can go crazy micro-focus adjusting their lenses if they don't have the setup correct. And this dot focus, from reading the info on the provided link, doesn't seem to work as well for Pentax cameras as it would on the Nikon system apparently.. since they have arrows to tell if the camera thinks the focused object is in front or behind. We'd have to eyeball it since we don't have those arrows.

At which point I think I'd rather do the ruler at 45 degrees test.. but one must ensure they keep the camera at the proper height and distance away (there is a site that has the calcuator and steps.. but it alludes me at the moment) otherwise they risk faulty results. I've experienced this first hand where the lens is apparently back focusing, I adjust with the ruler to center. Test in the 'wild' it appears to be front focusing. Recalibrate with the ruler now it's off again.. sometimes front, sometimes back.. wait what?

In live view it will focus fine because its using a different focusing type... I typically use that as my baseline though with the timer set to minimize camera shake.


TL;DR: Ensure your system is setup properly to perform mf calibration or you risk going nuts.

What he said, times 10!

This is a bit of a late addition, but I feel I it is worthy of emphasis.

---------- Post added 13-09-15 at 00:09 ----------

I agree. DotTune is one of the better calibration methods out there, but anyone expecting PDAF to consistently match magnified LV is going to be disappointed. The precision of the PDAF system simply is not there, particularly for subjects with low contrast and/or complicated contours.

A few notes regarding PDAF focus tuning:

• Just because you see evidence of front/back focusing, don't jump to the conclusion that AF can be improved through in-camera tuning
• Adjustment based on a curved or slant focus target is possible, but ill-advised
• A high contrast focus target is essential
• Never tune based on hand-held tests
• Do not base your adjustment on a single measurement. DotTune is intrinsically good in that regard.
• When doing DotTune (or similar), a wide spread indicates a target with inadequate contrast, a lens with inadequate contrast/resolution, and/or inadequate light. If you have accounted for light and target, decrease the distance used for the test
• Even when properly "dialed in" using the focus fine adjust, a lens may still front/back focus at a different focal length (zooms) or subject distance (internal focus designs).
• Accurate fine tuning at 2 meters does not automatically translate to accuracy or precision at 20 meters. A good rule of thumb is to not expect the AF system to do a better job than manual focus using the optical viewfinder.

Steve

Speaking as one who has gone crazy, your comments are well taken. Sure, my K-5 focuses accurately - sometimes. But consistency (i.e. precision) is simply not there. All I can do is hope that my next body will be significantly better. At least, comments on the K-3 AF are encouraging.