Im Cheap--Do-it youself AFMA Contraption

Well, Yes, I will spend $6800 on a lens but refuse to pay $79 on a device to fine tune the autofocus of my camera lens combo. After carefully reading Neuro's tips on AFMA and Bryan's review of one of the two competing AFMA targets I figured that I could do just as good for less than 10 bucks. Now one of those production targets is made of cardboard (high quality of course) and one is made of plastic. Each have a ruler attached to them and each have a high resolution target, and one even has a targeting system designed to drive any obsessive/compulsive (which most of us are) over the edge.

One of the problems I had was that I am trying to AFMA a 500L, and the recommended distance from sensor plane to target is 25-50x the focal length, or the approximate distance you will be shooting from; this equates to 37.5 feet to 75 feet, or I figured the distance I normally shoot from with this lens is over 100 feet---Unfortunately, both of the production targets are too small to cover a significant portion of the frame or sensor field. So I thought a larger target would be more appropriate.

My second thought was the targeting system of the more expensive unit---I have to ask why? How many subjects are at exactly the right distance and exactly level and exactly perpindicular to to the sensor plane?---My guess would be all of them or none of them--depending on your perspective---So, regarding targeting level, distance etc, well close should be good enough.

Next, I needed a good, high resolution target and found a full resolution download on the internet and printed it off @ 8.5x11 in, in high resolution and on satin paper to limit reflection.

I used a couple of scrap boards from the garage (one for the base and one for the target) and I used an old quick release plate to screw on to the bottom so I could attach it to an old tripod.

Finally, I needed a long ruler (with high resolution print) to finish this beast.

Cost:
$2.29 for the ruler (metal, 3 ft long, fairly high resolution print)
$2.50 for two board feet of pine, if you don't have any laying around the garage
$1.00 for the print (ink and paper included)
I used an old lens plate, but you could just as easily use an inset nut for $.29

total cost: $5.79

here is the end result:

Attachment 160

Do you think it will work?

I haven't had a chance to test it yet, weather has been windy and cloudy this week, but When I do test it I will be sure and report out the results-----Note: I shot this hand held from an elevated angle, but even so, you can see that I need some AFMA with this camera lens combo.

Bob my initial thought is no, the target is too large. With the 500mm you are going to be focusing on a black area or a white area. There is not enough contrast in the center.

The ruler should be fine it is long enough, however I find on the lensalign it seems certain angles work better than others.

The other problem you need to counter is how to keep the target square to the lens. With the lensalign you have two holes that you line your target up with.

IMO I think it is important to keep everything as level and square as possible, not that your shooting will be that way but more so you have consistency each time you try and test your lens. You want to have repeatability in your method, so that when you set it up again you get the exact same results. I did allot of testing with my lenses, and the more I was consistent with set up and lighting the closer I was able to tune the lens and see what they were doing.

For me it was easier with the smaller lenses in the house. I could use the floor as a gauge and just measure up to the target and lens and set them at perfect level. Then do the eye thing and it was all square. It is a bit harder outside, but I was lucky, my 500mm has been dead on 0 on every body I have had.

One other tip, is control your light. If the light varies from one test to the next, so do your results. Outside it is a test you would want to do during midday when light isn't changing much.

Just my thoughts Bob. Looks like your headed the right direction.

Edit: Bob you mentioned covering the sensor field, you should only be using one sensor and using single point center.

I think it'll work fine.

FWIW, the LensAlign has an optional (and additional cost, of course) long ruler kit - that's for lenses like the 500/4, and includes a longer ruler (duh) and also a larger focus target.

Rick's point is correct, though. The larger target isn't necessary for the recommended setup, 25x focal length. At that distance, the focus target of the LensAlign will be exactly the same size, regardless on focal length. It is fine to cover the center AF point, which is all you need.

The large target is used when you want to calibrate at a longer distance, to simulate how you use the lens (if that's how you use the lens, that is).

Yes, Rick, the long ruler kit for the MkII comes with the bigger target, too. Not sure if it attaches magnetically like the Pro version we have, though.

John

I deleted that post, I went to lensalign's website to see what the deal was. It is indeed a larger target, and since it is a kit you can add to the mark II, it is most likely a magnetic plate just like ours.

For what it is worth this is lensalign's website:
http://michaeltapesdesign.com/lensalign.html

Thanks All for your comments. I finally had a chance to go out and test this thing this weekend and here is my assesment of this cheap alternative:

I took great care in insuring that the target and the camera were carefully aligned---Identical height, perpindicular, Level etc. I ultimtely used a laser level that would shoot across my backyard. I also took great care in making sure camera adjustments were correct, i.e. Wide open aperature (F4), ISO 100, MLU, remote triggered etc.

Using the 500L, I set up at a measured distance of 75 ft from the target. At this distance the target seemed to be almost perfect in that the black and white intersection was just enough to cover the center AF Point and still fully view the ruler along the left side of the frame. Since I was fairly sure my camera/lens combo was front focusing, I shot 10 images at 0, +5, +10 and +15 each. I then uploaded the images into LR for review.

Upon review, My front focusing suspicions were confirmed and this is where it became very difficult to judge. With a depth of field of about 11 inches it was very difficult to tell where the in-focus started and where it ended with any level of accuracy. I also noticed that over the 10 shots, the readings were not consistant so I went with what the majority of shots indicated. Ultimately, I decided that the +10 setting was the closest. Not happy with the lack of accuracy at 75 ft, I moved the camera to 37.5 feet and repeated the test. This became much easier with the depth of field cut to approximately 5 inches. AT this distance I was able to confidently narrow down the settings between +10 and +15. I again repeated the test at +10, +11, +12, +13, +14, +15 (10 Shots each and determined that +12 was my best set---but there was still considerable doubt.

One of my biggest problems wasn't the target but instead, the ruler. It just didn't seem to have enough detail to make assesments that I was confident of.

I haven't tested this with smaller lenses and narrower depth of fields yet, But with long lenses, I am skeptical of this devices accuracy simply because the ruler just doesn't have the detail to make solid judgements.

So, Unfortunately, I don't think my $5 device can provide the level accuracy that the $150 device can, but that it can get you close if you you don't have or are unwilling to spend $150.

I will probably spend the $150 to be confident in my lens accuracy. But, It was certainly worth the effort to explore this alternative.

Thanks again for all your comments.

Bob

Don't give up yet. Some of the problems you have with your set up you will have with the lensalign.

A couple of things, the large target on the lens align is 5 3/16" square. The farthest the edge of the ruler would be from your focus point would be 2 5/8". I think your ruler is to far from your target. That will exaggerate whether you are off if you are just a little bit out of square.

Next, forget testing at the longer distances. With the 500mm lens if it is dead on 37.5' you are going to be dead on at 100'.

Now for your results, it took me a while to really figure out how to read the ruler with the lensalign the easiest. Some times the results are obvious. But results will are never consistent. How I do it is find the farthest numbers out that are in readable focus both back and foward. I take at least 20 pics at a particular setting. I use 1 star for front focused, 1 for dead on and 1 for back focus. I also take several pictures in live view focused at 10x to give me a template of what it should look like. I use the template to decide the numbers I will be reading.

After I download the pics I put them up on full screen. Then I literally scroll through the pics judging them and assigning stars, rather quickly as I do not pick over each one. For example at +4 on my 300mm I came up with Front Focus 18 On Focus 17 Back Focus 18. This seemed to be the best average as I went higher to +5 and lower the results would shift accordingly. (keep in mind when I say front focus or back focus, the lens is probably still within tolerances and whatever you are shooting is falling inside your DOF).

Focus is never dead on each time, and if you try and determine the absoulute zero focus point it may drive you crazy. I found it extremely tedious to do it that way.

One other thing you might do, is set up a flash on your ruler remotely. If not a flash run a light out to it and put it where it lights the ruler.

So I see at least two things to change. Get your focus point closer to the ruler. Make sure you have enough light on the ruler. Possibly change to a white ruler if that doesn't do the trick.

Quote:

---

Originally Posted by HDNitehawk

So I see at least two things to change. Get your focus point closer to the ruler. Make sure you have enough light on the ruler. Possibly change to a white ruler if that doesn't do the trick.

---

HD, Thanks for the tips. I ran out of time this weekend, but will defintely try your suggestions before dropping the big bucs.

I'll let you know how it goes.

Thanks again

One more piece of info that might help you. The lensalign calculator says that you don't need the large ruler and target at 37.5'.

To give you an idea of what the framing looks like, this is what my lensalign looked like when set up with the 500mm at about 38' using the 7D.


http://farm8.staticflickr.com/7015/6...f0f2c65a_z.jpg
Minus 3_0005 by hdnitehawk01, on Flickr

To determine alignment I used the middle row of numbers at 200% magnification

Thanks HD---I'll keep that in mind

Bob, I used an even simpler method for my 50mm f1.4
I printed a long black rectangle on a white a4 size paper(Capitol I would do the same trick). I pasted it to a door which was level.
I aligned my tripod and camera at about 2m(7 feet) which I assumed was the most used distance of the lens. In other words, the distance which I though would be most important to my shooting style.
Obviously I didn't have a ruler so I couldn't easily spot if my lens was front focusing or back focusing. I knew that it had a front frocus issue so I just took shots from AFMA 0 to AFMA 20 and I compared the shots on 100% view on my computer. Up till this day it has worked perfectly for me and it only cost me 1 sheet of paper:p
My 24-105 f4L needed a +4 and my Sigma 50mm f1.4 needed +13. My 70-200 f4L IS was spot on. The 100mm L macro had a different story. It was spot on at 1:1 to 1:2 macro settings, however for portraits it needed a +4 AFMA. Since I rarely used AF during macro, I used +4 on that lens as well.

As far as I'm concerned the Lens Allign couldn't do a better job. I'm pretty confident that I'm no more off than +- 1 which for me isn't a problem at all. With the Lens Allign I'm pretty sure that your method could also cost you +- 1 so in my opinion save the 150$.

Jan

I'm pretty cheap too. Would suggest two possible changes. One, replace the ruler with another that has finer graduations. Two, change the angle the ruler sits at. By making it more vertical, the existing graduations would have finer resolution.

JRW

Realized that I ought to elaborate a bit more. Am familiar with using rulers to set up vision inspection camera systems when the actual parts aren't available to use. Have found that satin finish rulers are best. The graduations I prefer are engraved and then black filled with a low lustre pigment. Have also used the same rulers to adjust micro focus. Using live view and manual focus, I focus on one specific measurement line or a target aligned with the ruler depending on what I am setting up. Take a picture. Turn on auto focus and compare the results of a few pictures. This is faster to do when tethered as it is also possible to adjust the focus from the PC. If the results are not close enough to keep me happy, I also know which way to adjust and just record the amount of adjustment required for adjusting in the camera settings. One other problem I experienced once when I had the ruler further away, was a pixelation issue which caused grief in assessing the results. The ruler was too close to horizontal which caused the relative width of the graduations to drop below one pixel width on the sensor. The separation of the markings was also very small in the images. Changing the ruler angle increased the apparent width of the markings in the images, while at the same time the separation increased for better resolution of the focos. It might be a good idea for you to back calculate the pixel resolution in millimetres using your sensor specs along with the image circle dimensions for the lens combined with the optical angle of view. Pixel peeping on one of the markings in an image would also give you a different path to the same information with no math involved.

Quote:

---

Originally Posted by Sheiky

As far as I'm concerned the Lens Align couldn't do a better job. I'm pretty confident that I'm no more off than +- 1 which for me isn't a problem at all. With the Lens Align I'm pretty sure that your method could also cost you +- 1 so in my opinion save the 150$.

Jan

---

Jan

Your method will work, however it takes longer to get the same results. It is a method I used for a long time.

What I see as one of the real benefits of the ruler is that it tells you what your DOF is doing overall. I don't agree that the LensAllign would cost you +/- 1, I would agree your camera will vary that much though. You can be +1/-1 and you will still have quit a few pics in focus, just as if you were dead on. There will be two reasons for this, one is that your AF system isn't pin point accurate, when you move forward or back you end up with either shots that were a bit back focused at dead center coming in to sharp focus or vice/versa for front focus if you go the other way. The other reason is that depending on your lens, your DOF still covers the target.

I would also say, that with the 500mm the +/- 1 matters more than with many other lenses. The wider the lens the wider the DOF, and of course a wider DOF is more forgiving. With the 500mm if you are shooting birds or small wildlife it is a narrow DOF, the fine detail of an eye can be one of the most important aspects of a shot. Luckily, I have found that the big telephoto's are the easiest to get dead on of all the lenses. They are much more accurate and consistent than any other lens I own. If you are a little off one way or the other your keeper rate will start to fall.

I still think Bob's configuration will work. The only thing I see left is he hasn't incorporated a way to easily square and align the target. The lensalign uses two holes that you peep through to get alignment. I think there would be a way to overcome that, with a little surveying and carpentry know how.

Bob, one other note. If you decided to buy a lensalign, you wouldn't have to buy the $150 version unless you wanted to test at the longer distances. They also sell the ruler and face plate as add on's, so if you think you need them later you can order it separate.

Rick