Started Sep 27, 2017 | Discussions thread
 Forum Parent First Previous Next
 Flat view
52

I was asked if I would make a post about doing a micro focus adjustment on the EM1.

So……………..Here it is…………………..

Before getting into doing the actual alignment I want to mention a few things about PDAF.

Oh………also of note - when I describe how CDAF & PDAF work I am talking about SAF.

But………lets talk about CDAF for a second. When a camera focuses with CDAF it does this really fast back and forth focusing and from that determines focus by looking at contrast in the area covered by the autofocus point the user selected. Because it does this back and forth comparison it is very accurate, much more accurate than PDAF.

Now…….before I go on I want to preface this with I am giving a very simplistic explanation of PDAF, as I did CDAF. I don’t have the time or energy to go into a detailed explanation of either.

PDAF uses the difference between light phases from two points to determine distance to subject (it also uses it to determine direction and speed for CAF) and it moves the lens to that position. There is no double checking. It knows based on the position of the lens elements at what distance the lens should be focused at. So, when it determines the distance using the phase differences…..it focuses the lens to what it thinks is that distance.

Now…………why did I say “it focuses the lens to what it thinks is that distance”?

There are two things to talk about concerning that question. The first is the fact that PDAF doesn’t focus to the same distance each and every time. If you put a camera on a tripod and a target at lets say 10 feet and press the focus button, then defocus and press the focus button and then repeat this over and over……………..

Each and every time the camera focuses it will be at a slightly different distance. Just because of the nature of things and the algorithms companies develop that are not 100% perfect the camera will determine a slightly different distance each time. The first time it could focus at exactly 10 feet and the next time 10 feet and 1/8 of an inch. But this difference will typically be within a specific range based in part on how good the algorithms are (plus the quality of the lens).

After writing the next part I realized I needed to come back up here and explain something a bit more. This first part is not what we are correcting for when we do a micro focus adjustment. There is nothing we can do about this fluctuation in the distance the camera focuses. This is the downside of PDAF compared to CDAF, the upside being how fast it can change focus distance when in continuous autofocus because it doesn’t double check and it’s why it’s the preferred method of focusing for CAF.

The second part of the equation is actually physical. It is the gears and other mechanical things within the lens. When the camera says focus at 10 feet the lens will move to what it thinks is 10 feet. Because every manufactured item has some tolerance for each piece, there will be error. How much that error is will be different for each lens.

Wait…………….need to point something out here

This second part can also refer to the camera but really only applies to DSLR’s because they use a separate sensor for focus (error can be introduced if the sensors are not aligned properly). With the EM1 its all on the same sensor, so any error in the camera is highly unlikely. But, I will confess to not having a good enough understanding in sensor design to speak with 100% certainty that there can be no error from the camera side of the EM1.

So……….this lens error. The camera says focus to 10 feet and the lens moves to what it thinks is 10 feet but is actually 10 feet and 1/8 inch. Now, this has nothing to do with the first thing I discussed. In the first section we assumed the lens was perfect.

This lens error is what we are correcting for when we do a micro focus adjustment.

Basically……………….

We are measuring this error and programming into those algorithms the error for this lens. So now the camera knows that the lens will focus at 10 feet 1/8 inch when it says 10 feet, so it tells the lens to focus 9 feet 7/8 inch. Which in theory should have the lens actually focus at 10 feet 1/8 inch and the photo be in perfect focus.

Yes………………very simplified explanation of focusing and what we are adjusting when we do a micro focus adjustment.

Next are some random things I want to put out there before continuing with the actual micro focus adjustment.

The error can be different at different distances. This is why you should always perform the test at your typical shooting distance for that lens. I have come across a few blog post discussing micro focus adjustments that mention error in the lens does result in different amounts of error based on focus distance but error in the camera (talking DSLRs here) is constant with distance. Making the importance of doing this alignment at typical shooting distance even more important.

When doing zooms you have to do both ends. The way the EM1 works is it takes the setting at the wide end and and the one at the tele end to compute the amount of adjustment needed for focal lengths between the two. I have no knowledge as to how accurate this method is. I pretty much only shoot my 50-200, 50-500 at the max focal length and I haven’t really noticed any problems with my 14-54 (which I use all the focal lengths) but I haven’t shot much with it. Just make sure to do it at both ends, otherwise only the end you do will be accurate and other focal lengths could be really off.

This is also not a one and done thing. It needs repeated on a regular basis because gears wear and wires stretch. Not electrical wires, but some lenses use actual wire to move the lens elements, some use gears and some use a combination of wire and gears. The point is……it is mechanical error we are adjusting for and the mechanical parts will wear and will increase the amount of error. I do my lenses every October and May, well I check them and I will explain that a bit later on.

The only reason I was able to write this up is because I happen to have not deleted all the data when I adjusted my Sigma 50-500. This actually works out better than if I had data from my 150/2. The Sigma is not as well made and you can see this by what for my standards are large differences in distance focused. A graph from my 150/2 would be much much tighter. This one also has a few anomalies that I see from time to time, so it’s good thing here for a teaching item.

So………lets get to it……………..

I use the Focus Tune system (http://michaeltapesdesign.com/) to do my adjustments. I know there are going to be people jumping on here and saying that you don’t need it and you can do it with some homemade contraption. While I agree that you could get lucky and get it exact or pretty close, the results are not going to be consistent and I will demonstrate why. I also know people are going to come on here and say the price is to high. For the software and target it is under \$200 and that is seriously stupid cheap in the grand scheme of things. I spend over \$200 a month just in gas going places to shoot, I drive a Jeep and it doesn’t get the best mpg. That is just in my normal photography runs, doesn’t even include at least one major trip a month to someplace for photos. Right now I have 4 lenses that I micro adjust, that is less than an extra \$50 for each lens…………seriously people…………………..\$200 for the system is really cheap compared to a lot of peoples overall spending on photography………….

Using the systems gives me piece of mind and I know that my lenses are calibrated to the best they can achieve. When I lay on the ground at the edge of the swamp to photograph a baby gator, I have complete confidence that my lens will focus accurately even with a DoF that is 1/2 inch (I don’t want to spend time shooting and shooting because laying on the edge of the swamp to photograph a baby gator is not the safest place to be………………mama is around and watching and she is very protective). I also know when that rare moment happens and I finally see a bird catch and eat a baby gator (shot I am still after) that my gear will focus correctly and I will get the capture I am after. That piece of mind is worth way more than the \$200 I spent on the software………..which was several years ago………….I still use the software, but the money spent has long been forgotten.

I only have original EM1’s, so I can only tell you where it is in that menu……….I assume it’s located in the same place on the mk2.

With that………………………….

So………lets select Lens Data.

When I select it I see a listing of my lenses similar to this

1 150 F2.0
1 150 F2.0 EC14
1 150 F2.0 EC20

Since I have my 150/2 without a TC on the camera, the first one is highlighted white and all other choices are grayed out. If you have never done this before your only option will be to arrow over to the right, that will take you into the adjustment area. At first all the focus points will be highlighted and that is what you want, will explain in a minute. So, you press the ok button and that changes the display some. You can now use the D-pad to move a scale on the right side from +20 to -20, this is where you set the adjustment.

When I first did this on my EM1 my OCD forced me to test every single focus point. I took an entire weekend doing this and determined that it’s a huge waste of time. I came up with a few differences of +/-1, but the time involved was just not worth it and I am not positive my small differences were not due to my error. I spent the next few times randomly testing different points (because my OCD forced me to at least check a few) each time I did this and now……..I just use the all points as I said above. What the all points does is use the center point for the testing and applies the setting you use to all the focus points. Just a faster way of setting them all the same. But the option is there to test each point and make a different setting for each. I have just not found a need for this. Maybe on a DSLR where the sensors (remember they use different sensors for image and focus) are not perfectly parallel this would be needed. But with all on one sensor in the EM1, I have decided it is not needed and have not seen any evidence to change my mind. I should note here that I use all my focus points. I shoot for composition and hate the focus and recompose method, so I do move the focus point around when shooting. Even doing this I have not seen evidence that one focus point is worse off than another.

Well………………..we now have established how to get into the adjustment area. Now to do the actual test.

First I setup my camera on the most sturdy tripod I have, it’s a holdover from my film days so it is heavy and sturdy. Then I put the target on another sturdy tripod and set the distance, this is where personal preference comes into play. Some suggest 25x the focal length and others suggest 50x, I think you should set the distance to what you typically shoot the lens at. In my first times I did test at 50x and again at 25x and again close to minimum focus distance. I did find some difference, which gives evidence to what I said above about error in lenses being different based on distance to subject. But the difference has always been pretty minor, within +/-1. So I now do 35x and test at minimum focus distance since I shoot a lot of close up stuff with my telephoto lenses. You can actually have multiple settings for the same lens and if I ever find a significant difference from 35x and minimum focus I would setup one for each. I would just have to go in and tell the camera which to use. Would not be a huge issue since I know when I am going to use the camera for close up stuff.

The focus tune system has a method for setting it up (not going into all that because they have videos that explain it all). But I do want to mention a few things. The first is that their method ensures the camera is perfectly parallel to the target and ruler. If you are not perfectly parallel you are introducing problems into the testing and will not get accurate results. The system also has you shooting a flat target, not a slanted ruler like most doing a homemade setup use. The advantage here is that if you shoot a slanted ruler you have no control over where the camera actually focuses. By that I mean it could use the top part of the focus point or the bottom or the dead center and that will change how the image looks as to where it focused. By using a flat target that is perfectly centered with the zero point of the slanted ruler you avoid any issue of where the focus point actually focused. One other thing about shooting a ruler, getting the center of the focus point at exactly the zero point is going to be next to impossible.

I am sorry, I just can’t see how a homemade setup or shooting a brick wall is going to be more accurate than this system. Sure it can be done but that is hoping for good luck and that you got everything perfectly aligned. Plus, \$200 really is stupid cheap if you care about having your gear perform at its optimal ability.

Now that you have all of that setup, it’s time to start shooting………………

The focus tune system has you shooting 5 shots at -20, -15, -10, -5, 0, +5, +10, +15, +20. Why 5 shots at each setting? Because of the variable focusing I talked about way back at the beginning. I have at times shot 10 photos at each setting but have found not much difference, so I save time and just do 5 shots. After you shoot that series of photos you upload them into the software and you get a graph similar to this.

The green dots are where the software determined each photo focused at. I have no idea what the values on the left represent as far as distance focused. I am sure an explanation is available, just haven’t researched it because it really doesn’t matter. It also always uses 400 as the extreme (will see that in the next graph), so it’s relative to the amount of error detected in test run. But looking at this graph you can see how the shots vary in the distance that it actually focused. The software provides these nice overlays on the photos with a mask so you can manually see how it focused each time. Below you will find a photo that has the zero line in nice sharp focus, followed by one where it back focused. The actual photos without the mask are also available if you wanted to look at them. I looked at this stuff carefully the first few times I did this, because I have OCD and not knowing drives me crazy. Now, I just run the test and ignore all this stuff…………….I trust the software.

Using those masked photos, which I am sure is based on contrast, it determines the focus distances that it displays on the graph. Something else I am sure jumped out at you is the +20 results. I have see this now and again and it can happen at either +20 or -20. I personally think it has something to do with the software in the camera and is an anomaly. I now just ignore it unless it looks like the actual adjustment is going to be between +15 and +20. If that is the case (I have one lens with ec20 that is adjusted to +17) I will re-shoot the +20 and it usually is fine the 2nd go around.

Finally, the blue diamond is the average for that set of shots. You see………….when we adjust the lens we can’t get it to focus perfectly each time because of the natural problems with PDAF. What we are trying to do is to adjust the lens so that the blue diamond is on the zero line. In other words, we want that natural randomness to fall as close to zero as possible……….giving us the most shots possible within the zone of sharpness.

Now, we see the line is passing zero somewhere between -5 and 0. So we go out and shoot 5 shots at -5, -4, -3, -2, -1, 0. This gives us something like this.

It has been to long and I don’t remember what I decided about the shots at 0. Can’t remember if I re-shot or ignored it, all I have are these screen shots I did (which was for doing a post about micro adjustments that I am just now getting to). Play along with me and lets say the line continued to go towards back focusing for the 0 shots, which is what I think I did. We have the -1 and -2 shots that are pretty close. The -1 is much tighter but that could be pure random chance or something I caused. Which this is a good time to talk about two things.

I know people who say they just setup a ruler and shoot one shot at each setting and look at them on the computer (because they don’t understand the randomness of PDAF focusing they think this is an acceptable test). So, lets say you did that for the -5 to 0 shots and you ended up with the green dot at the top for each setting. To you, the shots from -4, -3, and 0 probably all look pretty similar and the -1 & -2 shots look the worse. So what would you set it to, or would you re-shoot them all and this time maybe end up with all the bottom green dots. Now they all look bad except the -1 setting. I hope some are starting to see where the software really helps in all this. It will look at the photos and help figure out the average for each setting. If you go back and look at both graphs, you will see that the spread (randomness) for each setting is pretty close. Sure there are a few outliers, but overall that random distance I talked about at the very beginning is pretty close to the same for each setting. I also really wish I still had all the photos from the testing to show just hard it is to see a difference between most of these shots. This post is taking me a lot of time to write as is, without finding the time to also re-shoot samples……………guess you have to believe me.

The second thing is I have found that lenses perform a bit different as far as the randomness of distance focused if they are going from far to near or near to far when gaining focus. For this reason I always try to remember to defocus the same direction between each shot. I tend to defocus moving the focus point towards me, so the lens has to move from closer to farther away when it focuses. Sometimes I get lazy and just turn the focus ring without thinking and that could be what caused some of the randomness in the shots. But there is still one step left to help make sure we got the right setting. I have also been doing this long enough to know if I should re-shoot or not based on what I am seeing.

The software has a function called a focus consistency test. For that you shoot a series of photos all at the same setting, I like to do 50-100 shots. When you import them into the software and run the test you can group them into 5 shots clusters or 10 shot clusters (I run both if I shot 100 shots). What this does is give you a chart showing how consistent the camera focused at the correct distance. When I say I do my lenses twice a year, this is what I do. I run a focus consistency test and if it all looks good I move along. This test is simple and fast and I can do all my lenses with all combos of TC’s in a few hours, compared to about 20 minutes for each lens (40 for zooms). Oh, that 20 minutes is for the 150/2, and another 20 minutes for the ec14 and another 20 minutes for the ec20 (so an hour for a prime and two hours for a zoom if doing all possible combinations). For the focus consistency test I just sit outside and shoot all the shots for all the combos and then come inside and evaluate. Typically I use a different memory card for each lens to help me not get to confused.

Ok, so I ran the test for the Bigma and this is what I got. The first one at the -2 setting and the second one is the -1 setting.

I went with the -2 setting because it had the more consistent looking graph and was closer to the zero line. FYI, so far it seems to be working well based on photos I have taken.

I want to point out that when doing my other 4/3 lenses the shot patterns are tighter. This is my only Sigma 4/3 lens and I can only guess that it’s a quality control problem with it. The Bigma is not a high quality lens like the HG or SHG Olympus lenses. So they should have less variability due to tighter tolerances in the factory. My two HG lenses are pretty close to each other in spread pattern, while my 150/2 is the tightest of the bunch. Doing this really does point out how tighter tolerances can make a big difference in lens performance.

That is basically why you should do a micro focus adjustment and how I go about doing it. Now………………I guess I should mention a few things about the EM1’s.

When you use a 4/3 lens on any EM1, the camera will only use PDAF. That even applies to lenses like the ZD 14-54 ii that is CDAF optimized. For the original EM1, Olympus has come out and said that when using CAF the EM1 only uses PDAF…………regardless of lens used. That means that when you have an Olympus 300mm f4.0 on your EM1, the camera is only using PDAF and doesn’t use CDAF when you select CAF. This does mean that there could be focusing error when using m4/3 lenses in CAF on an original EM1. I don’t own any m4/3 lenses that I use for action and have not tested this personally. But I can promise you if I did, I would be checking this…………..more on that in minute.

For the EM1mk2, Olympus is being very tight lipped about how their focusing really works. I personally believe that they still basically only use PDAF for CAF. I hear some say that it uses PDAF but checks focus accuracy with CDAF, but I find that very hard to believe. The idea that a lens is moving it’s elements to keep up with a subject coming towards the camera I understand and can rationalize how it works. What I can’t fathom is to use CDAF to check accuracy it would have to also be moving back and forth along some distance it believes the subject is located while also moving the focus closer to the camera at the same speed as the subject and that just doesn’t seem realistic to me. But, as I said…………I don’t own an EM1mk2 so I can only speculate on how it focuses with m4/3 lenses. But………………………………….

A good friend who’s work I admire and skill I trust, does have a mk2 (he is also the one who turned me on to the focus tune system). He tells me he does see a difference in keeper % when shooting in CAF now that he adjusted his MZ 300mm f4.0 Pro and MZ 40-150 f2.8 Pro lenses. He believes that with the mk2 it is still needs to be at least checked and if needed adjusted. There does seem to be a problem and the camera doesn’t report the micro adjustment in exif. Oh, yes………The exif data does include the micro focus adjustment value. But there is some problem with the mk2 and it doesn’t report that info, even tho it actually does use the adjustment. This problem is only with m4/3 lenses, with 4/3 lenses it seems to always report the adjustment. Again…………this is not my personal experience but from someone I know and trust, so I am pretty confident in the information.

……………………………..wonder why that is?

For those using the 4/3 TC’s. I personally believe when people talk about the ec20 being soft and unusable it’s because they didn’t do a micro focus adjustment. My experience has been that the lens needs a minor adjustment, the ec14 more and the ec20 even more. You can go to my Flickr and see a lot of full resolution samples of the ec20 on my lenses (all shot wide-open) if you are curious. The 4/3 TC’s are remarkable and perform amazingly well when you take the time to align your gear properly. It also explains why some say they need stopped down. Stopping down increases DoF, which going down a stop could be just enough to put the 0 line into sharp focus.

I do want to mention something here about using not just 4/3 lenses but PDAF in general (for any DSLR users reading this). I typically shoot in SAF, even when photographing a bird catching fish. I also shoot in burst to try and capture the perfect shot. When shooting a bird catching a snake (which I did, took like 10 minutes for him to eat it) I shoot in short burst for several reason. The first is so I don’t fill my buffer, that way I have room if something super cool happens. The second is because I know the camera will focus at a slightly different location each time. But doing short burst and forcing the camera to refocus each time I increase my odds of getting the sharpest photo possible.

I know that was a long read and a lot of information and I really do wonder how many actually read the entire thing. I hope some people learned something and that it helps them in the future.

I am willing to answer any questions that I can about the process……………………

Ronnie

Phocal's gear list:Phocal's gear list
Olympus Zuiko Digital ED 150mm 1:2.0 Olympus M.Zuiko 300mm F4 IS Pro Olympus OM-D E-M5 Olympus E-M1 Olympus M.Zuiko Digital 17mm 1:2.8 Pancake +6 more
Olympus E-M1 Sigma 50-500mm F4.5-6.3 DG OS HSM
If you believe there are incorrect tags, please send us this post using our feedback form.
Complain
 Forum Parent First Previous Next
 Flat view
Post ()
 Forum Parent First Previous Next
Keyboard shortcuts: