[UncleVanya]

Here's my feedback for your latest tweaked instructions. And before that my thanks - Thank you!

How to Use LensAlign MkII Focus Calibration System
Once you get the kit setup per the included directions, you'll need to know how far apart initially to set your camera and target. The basic distance formula is focal length x 25, in millimeters. Here's a cheat sheet.

Any idea how much tolerance there is for going longer or shorter?

Initial Setup
* Work in a constantly lit area.

Did you mean consistently or constantly? Is there any guide as to how much light is needed? Some cameras struggle with autofocus in low light and many lenses are slow - is there a guide? For example on my K50 in my latest tests I was shooting 1/10 a second at ISO 100 (18mm f/3.5 on my 18-135 lens)

* Put camera into manual mode.

Why not TAv?

Ensure Straight and Parallel Targeting
* Aim your camera at the middle, largest dial on the cards left side. Sometimes, if shooting from a long distance, you might try aiming for the right circle instead.

Since parallax will be worse at the shortest distances why would this be more likely something to do when shooting long?

* It might be easier to totally defocus, then aim for the center.

Between adjustments or just during setup?

* Take a picture.
* Check the picture at 100%.

On Computer or camera LCD?

Seems Easier, but Not Auto Focus Adjustments
Seems Hard, but Not Recommended Auto Focus Adjustments

The way these are labeled is confusing right now. Perhaps you might consider labeling as Method 1 and 2 and then post a brief note about each one. For example...

Method 1: Please note that this method only appears easier, typically this involves many trips back and forth from camera to computer.
Method 2: Please note that this method looks more difficult but reduces trips back and forth between the camera and the computer.

Also clarifying the global and individual adjustment options for focus might be helpful. For example - do they overlap and stack? Does -10 to all and -10 to a specific lens = -20 when using that lens?

[UncleVanya]

I repeated my tests at 18mm, 35mm and 135mm with my 18-135. A few things came out in the wash. First the 18mm results looked best at +6, the 50mm looked best at +5 or +6, and the 135 needed +8... that sucked to find out. I went back and looked and +7 seems like the best compromise based on the results.

Another thing that came up is that focus adjustments did not seem to make a difference on the Contrast Detection (Live View) method of autofocus. Is this normal? Additionally and maybe a bit weird, the exposure of the automatic metering (forgot to go to manual) gave a brighter picture (better exposed in truth) using the LiveView - I had my viewfinder metering set to full automatic matrix metering. Lastly when I tried to see what the impact of Apply ALL and Apply ONE was it was odd. I tried setting +10 on All and -10 on ONE but that did NOT result in the same image as +/- 0, it was slightly farther backshifted than the +/- 0 results. Consistently the ALL seemed to undercorrect compared to the ONE setting.

One more thing - the lighting is not great in my room - 95 watt halogen flood reflecting off the ceiling, some natural light and that's about it. The 135 focal length gave focus confirmation on both CDAF and PDAF using the small target but the results were terrible. The focus was not even close. Switching to the large target was a necessity!

[comprock]

Here's my feedback for your latest tweaked instructions. And before that my thanks - Thank you!



Any idea how much tolerance there is for going longer or shorter?




Did you mean consistently or constantly? Is there any guide as to how much light is needed? Some cameras struggle with autofocus in low light and many lenses are slow - is there a guide? For example on my K50 in my latest tests I was shooting 1/10 a second at ISO 100 (18mm f/3.5 on my 18-135 lens)




Why not TAv?


Since parallax will be worse at the shortest distances why would this be more likely something to do when shooting long?



Between adjustments or just during setup?



On Computer or camera LCD?



The way these are labeled is confusing right now. Perhaps you might consider labeling as Method 1 and 2 and then post a brief note about each one. For example...

Method 1: Please note that this method only appears easier, typically this involves many trips back and forth from camera to computer.
Method 2: Please note that this method looks more difficult but reduces trips back and forth between the camera and the computer.

Also clarifying the global and individual adjustment options for focus might be helpful. For example - do they overlap and stack? Does -10 to all and -10 to a specific lens = -20 when using that lens?

I've updated the how to at https://www.pentaxforums.com/forums/58-troubleshooting-beginner-help/249871-l...ml#post2669286 per your wonderful feedback. Any more revisions suggested? Others are welcome to critique.

---------- Post added Dec 12th, 2014 at 07:23 ----------

I repeated my tests at 18mm, 35mm and 135mm with my 18-135. A few things came out in the wash. First the 18mm results looked best at +6, the 50mm looked best at +5 or +6, and the 135 needed +8... that sucked to find out. I went back and looked and +7 seems like the best compromise based on the results.

Another thing that came up is that focus adjustments did not seem to make a difference on the Contrast Detection (Live View) method of autofocus. Is this normal? Additionally and maybe a bit weird, the exposure of the automatic metering (forgot to go to manual) gave a brighter picture (better exposed in truth) using the LiveView - I had my viewfinder metering set to full automatic matrix metering. Lastly when I tried to see what the impact of Apply ALL and Apply ONE was it was odd. I tried setting +10 on All and -10 on ONE but that did NOT result in the same image as +/- 0, it was slightly farther backshifted than the +/- 0 results. Consistently the ALL seemed to undercorrect compared to the ONE setting.

One more thing - the lighting is not great in my room - 95 watt halogen flood reflecting off the ceiling, some natural light and that's about it. The 135 focal length gave focus confirmation on both CDAF and PDAF using the small target but the results were terrible. The focus was not even close. Switching to the large target was a necessity!

@UncleVanya, I've not dealt with the Contrast Detection yet. Didn't seem to need it so far. Regarding metering, I'm personally trying out spot more with my K5 ii for the most part.

[UncleVanya]

The new write-up is much cleaner and I suspect easier for newbie's to read. I like the changes you made to my suggestions and think it is a great improvement.

---------- Post added 12-12-14 at 05:12 PM ----------

The standard test procedure for measuring the focusing
accuracy of any lens using the LensAlign system is to
position the focus target at a distance from the camera
equal to 25X the focal of the test lens. This can be simply
thought of as 8 feet per 100mm of lens focal length (so
test a 100mm lens at 8 feet, a 75 mm lens at 6 feet and a
50 mm lens at 4 feet).. Measure the approximate distance
from the camera body to the front of the LensAlign focus
target. It is not critical that this distance be precise. Within
a few inches is fine at close distances, and within 1 foot is
fine for distances beyond 15 feet.

From the user manual.

Also they recommend monochrome setting for this test and they recommend shooting against a black or dark background. Additionally - the exposure recommended is based on ISO 100-200 but they say you can increase if needed to eliminate potential shake.

Camera Exposure
The exposure should be as bright as it can be without
blowing out the Whites of the Display Ruler. This is best
accomplished, by observing the “blinkies” (highlight
warning) of test exposures, and adjusting the exposure
until the blinkies on the ruler section of the image just dis
-
appear. There should be no concern if other parts of the
image are blown out. We want the ruler to be optimally
exposed as brightly as it can be.
We recommend that you shoot in manual mode so
that each LensAlign test image has the exact same
exposure.

Lastly if the lens is not sharp enough wide open the instructions suggest stopping down 1 to 2 stops until it is sharp enough to determine focus. This will come at the expense of a greater depth of field (DOF) which could be a problem if the DOF is too large.

[narual]

it's a pity the focustune software isn't transferrable. It makes this all a little bit less tedious.

[mee]

Is the entire device supposed to be tilted back like that? right side is further back than left side..

[UncleVanya]

Is the entire device supposed to be tilted back like that? right side is further back than left side..

It was aligned using the method specified in the tool. I don't see what you are seeing, but errors might have crept in.

[narual]

Af fine tune is not for live view. Live view is much more accurate. Just slower. If you use live view at all for anything aside from getting the device aligned with the camera, you're wasting your time and making an already tedious process even worse.

There are pretty good, easy to follow directions on the lens align forums.

Some common problems the inventor points out with people's setups when they're getting in consistant results (some of this is more dpecific to focus tune software):

• Steady stable tripod required. Lens align can be on a light stand - it doesn't weigh much. Ideally Indoors so wind isn't messing with the target. (Random tip: the screw thread on top of a typical lampshade is the same as a tripod screw - so you can put it on top of a nice, heavy lamp, which should be very stable)
• They both have to be level. If either one is tilted you're wasting your time. (You need to set up with the camera focused roughly on the center of the rightmost circle, then look through the center circles on the back and adjust til you're looking at the center of the camera lens (this part gives me a huge headache every time). Then put the large target on)
• The targets must be well lit, and evenly lit. Shadows will give inconsistent results, as will noise from high ISO.
• Don't move the camera while you're adjusting. Tiny movements between shots don't matter too much, but use a light touch on the camera (especially where you have to use the AF switch on the body).
• Use a remote to take the shot. Mirror lockup mode isn't a bad idea either.
• Don't be too close or too far away. You should be about 25x the focal length. (Shorthand for Americans - focal length in inches is close enough til you get to 200 or so.) It doesn't have to be exact. It needs to be closer than the lenses' infinity distance, further than minimum focus distance, and close enough that the pattern is large enough to work with. If you can't do it at the recommended length for some reason, a little bit further is probably better than a little bit closer because focusing errors are magnified over distance. Exceptions would be lenses you use at specific lengths, like a studio portrait lens or a macro lens - those you'd use at that specific distance, and in the case of the macro lens, use the macro marked hole for alignment. I think it's the one closest to the ruler)
• Using the software, make sure to select the 5 red dots on the first photo, and check it against the last one - that will let you know if you moved anything more than the software could correct for, as well as ensuring it knows where to look for results.
• Oh, and my personal screwup... Make sure the blasted thing is set to jpg with no in camera sharpening applied. I did a whole set in dng on 2 occasions and one with jpg sharpening turned up all the way from messing with it in the past and forgetting to turn it back.

[UncleVanya]

Thanks for clarifying a few things. Most of this is what I already was doing but turning off jpg sharpening was a new trick. I also had not had confirmation about CDAF not using adjustments so that is helpful to know also.

Af fine tune is not for live view. Live view is much more accurate. Just slower. If you use live view at all for anything aside from getting the device aligned with the camera, you're wasting your time and making an already tedious process even worse.

There are pretty good, easy to follow directions on the lens align forums.

Some common problems the inventor points out with people's setups when they're getting in consistant results (some of this is more dpecific to focus tune software):
Steady stable tripod required. Lens align can be on a light stand - it doesn't weigh much. Ideally Indoors so wind isn't messing with the target. (Random tip: the screw thread on top of a typical lampshade is the same as a tripod screw - so you can put it on top of a nice, heavy lamp, which should be very stable)
They both have to be level. If either one is tilted you're wasting your time. (You need to set up with the camera focused roughly on the center of the rightmost circle, then look through the center circles on the back and adjust til you're looking at the center of the camera lens. Then out the large target on)
The targets must be well lit, and evenly lit. Shadows will give inconsistent results, as will noise from high ISO.
Don't move the camera while you're adjusting. Tiny movements between shots don't matter too much, but use a light touch on the camera (especially where you have to use the nd switch on the body).
Use a remote to take the shot. Mirror lockup mode isn't a bad idea either.
Don't be too close or too far away. You should be about 25x the focal length. (Shorthand for Americans - focal length in inches is close enough til you get to 200 or so.) It doesn't have to be exact. It needs to be closer than the lenses infinity distance, fort her than minimum focus distance, and close enough that the pattern is large enough to work with. A little bit further is better because focusing errors are magnified over distance. Exceptions would be lenses you use at specific lengths, like a studio portrait lens or a macro lens - those you'd use at that specific distance, and in the case of the macro lens, use the macro marked hole for alignment. I think it's the one closest to the ruler)
Using the software, make sure to select the 5 red dots on the first target and check it against the last one - that will let you know if you moved anything more than the software could correct for, as well as ensuring it knows where to look for results.

Oh, and my personal screwup... Make sure the blasted thing is set to jpg with no in camera sharpening applied. I did a whole set in dng on 2 occasions and one with jpg sharpening turned up all the way from messing with it in the past and forgetting to turn it back.

[narual]

Thanks for clarifying a few things. Most of this is what I already was doing but turning off jpg sharpening was a new trick. I also had not had confirmation about CDAF not using adjustments so that is helpful to know also.

Also the camera should be set to a neutral setting. Standard or neutral or natural, I forget what it's called on Pentax - his instructions are using some Canon terminology and my camera isn't in front of me to check it.

The forum link is here, but to save some time, here's his list of "Best Practices" which includes some canon/nikon specific issues that don't apply with Pentax.

• If Outdoors, use Aperture Priority Exposure - OK if lighting varies slightly - watch for glare and shadows
• If lighting is constant artificial source (indoors), use manual exposure, manual ISO
• Set Picture Style to Standard (not Vivid, Landscape, etc.)
• Large-Fine Camera-shot JPEG images (not RAW conversions)
• Single Center point AF (not Spot AF on Canon)
• Single shot (not continuous)
• Fixed White Balance (not Auto)
• Shoot Wide open (in some cases it is best to shoot 1/3 to 1 stop down, especially with lenses 1.4 or faster)
• All stabilization systems OFF (Canon IS, Nikon VR, etc)
• Shake avoidance techniques
• AFA settings (AF MicroAdjust or AF Fine-Tune) should be made to the "per lens" or "saved value" setting, not the "default" or "all lenses" setting
• de-focus between shots slightly, in the same direction (towards "closer", not infinity....do not de-focus to extremes)
• Do not shoot in Live View mode
• Do not use rear AF button on Canon cameras (Canon bug).
• You should use a cable release or self timer.
• Use highest exposure without causing the ruler or target to clip (watch the blinkies on the rear LCD to determine exposure)
• ISO up to 1600 if needed (better to use ISO 1600 than to risk camera shake with lower shutter speed)
• DX Crop mode on Nikon (just to keep file size down), but FX/DX does not affect outcome
• Position LensAlign and Camera using True Parallel Alignment, then mount Large Gen 3 Focus Target
• Viewfinder eyepiece closed
• Shutter Delay Mode (Nikon)
• Self-Timer (Canon)
• Cable release

He also suggests broad tuning before fine tuning -- take 5-10 photos each at -10,-5,0,5,10, then run them through the software (or without it, just eyeball them, I guess) and that should give you a good idea of where to fine tune (i.e., if your sharpest images are at -5, and -10 is worse than 0, you should run another set at something like -6, -4,-3, -2, and -1.

The FocusTune software is nice because it does the comparisons in seconds, and it's more accurate. Like, here's my result for my 77mm ltd, which shows that my best adjustment is probably -4/5 and I need to do more testing (shot deviation should stay under 100 - the lower, the better.) It's too bad it's fairly expensive and the license is only good for 2 computers.

[UncleVanya]

Why close the viewfinder? That's really annoying in a workflow. Why also do they say to not focus towards infinity? I usually defocus in opposite directions each time.

[narual]

Why close the viewfinder? That's really annoying in a workflow. Why also do they say to not focus towards infinity? I usually defocus in opposite directions each time.

Not sure why it would be annoying when you don't have any reason to look through it. The reason to block it off is because you're not looking through it, and when you're not looking through it, any light in the room can get into it. That can affect your exposure settings when you're not in manual, and it can affect your image, too ('light leaking'). Every dslr camera comes with a little plastic cap to cover it with. I'd imagine that if you don't have a lot of light streaming into your eyepiece, it probably won't matter much. But it's something to consider if you're getting poor results.
Cover Your Eyepiece - Digital Photography School

Not focusing towards infinity - I'd imagine because you want to be coming from the direction with less depth of field (remember the area in focus is 1/3 in front and 2/3 past the actual focal point, not dead center), but thats just a guess. I'd go with what the inventor of the product suggests.

Doing it in the same direction each time reduces variability. These are tiny adjustments - they don't need huge focus changes on your part, and doing those will give you less consistency (I think this is why some people autofocus, then slightly defocus and then autofocus again - they get a more accurate result when the lens is close to correct than when it starts from farther away). I know for sure before I read the post with that info, I was getting much less consistency, doing what you were doing (but maybe worse. I was doing 3 shots from all the way to the left, 3 from all the way to the right, and 2 each from a close distance in each direction. It was very frustrating, which was why I ended up looking on the forum and finding that information.

[UncleVanya]

Not sure why it would be annoying when you don't have any reason to look through it. The reason to block it off is because you're not looking through it, and when you're not looking through it, any light in the room can get into it. That can affect your exposure settings when you're not in manual, and it can affect your image, too ('light leaking').

I am in manual so no exposure change - and in a well lit room at reasonable shutter speeds I have a hard time thinking this is more than just extra work. Also I AM looking through it before each series to ensure that it hasn't done something stupid like focus crazy. I have seen the camera focus (even with center spot focus) poorly once in a while so I make sure it looks rational before I push the remote button.

Not focusing towards infinity - I'd imagine because you want to be coming from the direction with less depth of field (remember the area in focus is 1/3 in front and 2/3 past the actual focal point, not dead center), but thats just a guess. I'd go with what the inventor of the product suggests.

If this were true the entire test would be void. You only get this result at a specific distance from the lens and it is longer than the recommended distances. Think about it - you would not be able to adjust the focus for back and front focus by making them evenly distributed if it was supposed to be 1/3 and 2/3...

Doing it in the same direction each time reduces variability. These are tiny adjustments - they don't need huge focus changes on your part, and doing those will give you less consistency (I think this is why some people autofocus, then slightly defocus and then autofocus again - they get a more accurate result when the lens is close to correct than when it starts from farther away). I know for sure before I read the post with that info, I was getting much less consistency, doing what you were doing (but maybe worse. I was doing 3 shots from all the way to the left, 3 from all the way to the right, and 2 each from a close distance in each direction. It was very frustrating, which was why I ended up looking on the forum and finding that information.

Hummm... I think that begs the question, "Am I adjusting to focus for this test or the real world"? In the real world my focus points will not be predetermined and the last point of focus may be longer or shorter than the current needed one. To my mind - randomizing or using a pattern for this makes more sense so that both under and over focused images are adjusted correctly.

[narual]

I am in manual so no exposure change - and in a well lit room at reasonable shutter speeds I have a hard time thinking this is more than just extra work. Also I AM looking through it before each series to ensure that it hasn't done something stupid like focus crazy. I have seen the camera focus (even with center spot focus) poorly once in a while so I make sure it looks rational before I push the remote button.

If this were true the entire test would be void. You only get this result at a specific distance from the lens and it is longer than the recommended distances. Think about it - you would not be able to adjust the focus for back and front focus by making them evenly distributed if it was supposed to be 1/3 and 2/3...


Hummm... I think that begs the question, "Am I adjusting to focus for this test or the real world"? In the real world my focus points will not be predetermined and the last point of focus may be longer or shorter than the current needed one. To my mind - randomizing or using a pattern for this makes more sense so that both under and over focused images are adjusted correctly.

You don't have to convince me, man. The only person who's going to be affected by how your results turn out is you. The inventor has gone over this stuff repeatedly and posted those tips on getting the best results. Do it, don't do it, whatever. If you're not using the focus tune software, the little things are probably not going to matter anyway because you're just eyeballing it.

If you're getting bad focusing with your remote control, use the AF button on the camera body and don't let go of it til it's actually done focusing. Sometimes it overcorrects and goes back and fixes itself a moment later.

The 1/3 - 2/3 stuff is a rough number -- my apologies. The specific amount can vary, and I should have clarified that. But if you focus specifically to X feet, a larger part of the slice that's in focus is going to be behind your specific focus distance than in front of it. That's not opinion, and it's nothing to do with lens align or focus adjustments or anything - that's just how depth of field works. Play with a depth of field calculator. It's easier to see when you give yourself a larger depth of field, of course. Even just looking at the DOF scale on an old lens - notice the closer the distance, the closer together the numbers are on the lens barrel, and the further away you focus, the wider apart they get? See the bars on the aperture ring that show what the depth of field is? The distant part is always going to be larger than the closer part, though at wide enough aperture it's going to be so small a difference it's not really relevant.

As for "am I doing this for test, or for the real world" ... really? You're shooting at a focus chart. Of course you're doing it in a non-real-world environment. That doesn't mean that it's not going to fix problems with real-world focusing. When you look at an eye chart in the optometrists office and he's testing different corrections to see where you get the best vision, this isn't invalidated because you're looking at a chart instead of at a christmas tree with 12 tiny reindeer dancing in front of it, is it?

[UncleVanya]

It is true that a larger portion of the background should be in focus at certain distances - but at the distances we are using the chart that needs to be untrue or we are wasting time trying to even it out. The DOF calculator for a 50mm lens used here shows a fairly balanced result up until 5 meters... and if we shoot at 50 inches or so I'd suggest that the difference front to back is nearly zero.

The point is that our focus points used in the test are intended to be within the evenly split focus zone. Perfect - no, but thats the theory as I understand it.

Your point about reality is interesting. The point of the test is to get a baseline and adjust it to close to perfect and then shoot real world subjects as you would do so to see if the results are accurate. I can see that validation step as being sufficient to reduce the test variance to negligible importance - the important part being to defocus in some way to ensure that the adjustment made is being tried during the testing.

Thank you again for helping me through this mental process. I don't have the software so it is more subjective and less clear.

One more thing. Many of the instructions say to use the right most circle - but the instructions on the web from the manufacturer differ from this. I don't see how using the right most circle works given that you cannot ensure that the target is parallel unless you use the center one - what am I missing? (I do understand that you can align it and then put the large target on - that part I have no confusion with.)