Microadjustment

[Loswr]

The image you posted of the chart didn't include the whole thing (similiar to the spydercal) which was misleading at best. It look liked a printed chart. These tools seem to be the best solution. But I have found that when going shooting real life subjects, I still need to make adjustments, which kind of is crazy considering how much these tools cost.

The Spydercal has a level on it so it's simple to keep that and the camera parallell.

Sorry about the misleading pic - I was looking for an image that showed the thin DoF you can easily measure with a tool like that.

I think that while they're expensive for what they are (production costs can't be very high), they're cheaper than almost any Canon lens and a LOT cheaper than even the cheapest L lens. Like a good RAW converter, they're something that can benefit nearly every shot you take with your camera(s) and most lenses. Worth it, IMO, especially considering the less than adequate results you can get with the free equivalents to those tools (I, too, have spent time with a cut-and-fold paper DoF chart!).

I didn't notice the SpyderCal has a level - that would do the trick, as do the sight gates on the LensAlign. Both the SpyderCal and the new LensAlign MkII also fold flat, unlike my Pro version that takes up more space when stored.

 

[aldvan]

After reading and reading again all your posts, I spent a couple of days to setup a good microadjusting session. I opted for Jeffrey Friedl's method, that was well explained and didn't require to wait (and pay...) for a lensAlign to be shipped. The chart is well designed and the logic of the light grey indicators seems quite reasonable.
The lack of an instrumental system for a perfect alignment camera target required some additional time to spend leveling and measuring, but at last everything was perfectly aligned.
I started with the 1Ds3 and the 100-400, followed by the 5D2 and the same lens. At the beginning everything seemed working well, following well every step of the procedure. Frequently tested if the sensor was reading just the right target, turned off IS, fast shooting time, manually scanned from minimum distance to infinity before AF, camera on heavy tripod. Both cameras-lens seemed affected by a slight front-focusing corrected at -5 and -6.
Then, just for checking, I tried to set at -9 with the unlikely result to get again a front-focusing. Tried and tried again, getting similar results.
After reading an old article by Tom Jackson about the disalignment between the real sensor and the focus outlines in the viewfinder, I convinced my self that a lab procedure for microadjustment is risky and too exposed to minimal error factors.
So I reset everything and tomorrow I will go for what seems to me a more empirical and real world approach.
I will puth cameras and lenses on my tripod, aiming at the right distance for a wall plenty of contrasting and fine details. Then I will take 41 pictures focusing only with the central spot, adjusting from -20 to +20 and I will compare the results in LR3.4.
Quite rude but nor too time consuming, and it seems to me the exact procedure for what the MA system was designed...
Any opinion?

 

[Loswr]

Sorry, but to me it sounds like you tried the free method, and got what you paid for. Even the blog post you mention recommends LensAlign - there's a reason for that. For one thing, that one horizontal line is not optimal for some sensor point geometries (bear in mind the guy who developed that chart shoots Nikon, and they don't use the high-precision diagonal center AF point found in some higher-end Canon bodies).

Also, you plan to shoot just one shot for each adjustment, and possibly that's what you did the first time. That's not a good plan - AF is accurate within a range, so you need several shots at each setting. Personally, I do 8 - two from infinity, then two without refocusing, then two from MFD, then two more without refocusing. Given the requirement of many shots per setting, comparing sharpness in that 'wall test' will be quite a challenge.

 

[aldvan]

Neuroanatomist, I understand what you mean, but, at the same time, I'm beginning to understand also what Canon states in 1Ds owner manual: "Do this adjustment only if necessary. Note that doing this adjustment may prevent correct focusing from being achieved".
You told about taking 8 shots for each adjustment. That speaks to me about a random behavior statistically corrected, badly exposed to the risk of any random process... I don't know, now...
Since I haven't focusing problems, I'm seriously tempted to let everything as mother Canon did...

 

[Viggo]

I have microadjusted every lens, and although, before I found out how, it caused me many sleepless nights. BUT to see every picture not as sharp as it was with liveview-af or MF, is even worse. It takes some trial and error and if a lens is too off (like mye 300 f2,8 was) you need to get it hardware calibrated with Canon first, otherwise it won't help if you're at +20 or -20 with Ma.

What Canon do is solder one wire from one place to another in 4 steps. This is to move the POF a larger amount, and usually it helps a GREAT deal, but sometimes you find that your lens on your body is sharper in between those steps, and needing the MA to make it juuuust right. If you own a 2,8 lens or slower, it's less noticeable, but for me, who mostly use 1,2 and 1,4 lenses, this is the difference between no shot or the perfect shot.

Neuroanatomist: I agree with you on that note, compared to not getting what you paid for in a lens, that can easily be corrected, they are cheap. But considering it when you hold the cheap piece of plastic in your hand and think that you paid half a 17-40 for it is kinda strange. Luckily for me I can borrow one from work. And as soon as I get back, I'll take your advice and properly set it up, and adjust all my lenses one more, well, not the TS 17mm ;D

 

[Loswr]

You told about taking 8 shots for each adjustment. That speaks to me about a random behavior statistically corrected, badly exposed to the risk of any random process... I don't know, now...
Since I haven't focusing problems, I'm seriously tempted to let everything as mother Canon did...

That is exactly the way it works - any mechaincal (or biological, or physical) process has some degree of randomness. AF is no different. Not every shot will be perfectly focused, period. Precision tells you how close one shot is likely to be to the next - with a high-precision center AF point and f/2.8 or faster lenses, shots fall in a range of 1/3 of the depth of focus at max aperture, with slower lenses, the precision is within one depth of focus at max aperture. Accuracy is different than precision. AFMA corrects for accuracy, but during the process of calibration, you still need to account for precision - thus, the multiple shots. Less-than-ideal accuracy is most evident with a thin DoF (fast lenses shot wide open with fairly close subjects).

You may not know you have a problem unless you prolerly test for it...

Neuroanatomist, I understand what you mean, but, at the same time, I'm beginning to understand also what Canon states in 1Ds owner manual: "Do this adjustment only if necessary. Note that doing this adjustment may prevent correct focusing from being achieved".

Canon's advice is sound - I bet there are a lot of people out there who do the AFMA incorrectly, and end up in a worse place than no adjustment at all. Done properly, it's wonderful - pros used to send bodies + lenses to Canon for this sort of calibration (that was behind the genesis of Canon Professional Services, actually). Now, I can do it at home, my self, for less money and with faster turnaround. Almost all of my lenses have some amount of AFMA, and I can't imagine shooting with lenses like the 85mm f/1.2L II or 135mm f/2L without that feature.

 

[aldvan]

Thank you, neuroanatomist. I'll bite the bullet and I'll order the LensAlign, although it is a steal!

 

[aldvan]

Just one more question. To get rid of the pain of taking out the card and downloading the pictures, could it be a good idea to connect the camera by EOS utilities to a computer and check the focus directly on the monitor?

 

[Loswr]

Might work. I do find myself going back and forth to compare images, though. I'd suggest using tethered shooting to narrow the range, but still shooting (capturing) multiple shots 3-4 units on either side of what initially seems to be the correct setting.

 

[PeterJ]

After reading this thread I ordered a LensAlign MkII which finally turned up today. They posted quickly but mail from the USA to Australia seems to have been a bit slow lately. Anyway I thought I'd post my results after using it on my lenses with a 7D:

24-70mm f/2.8L, +5
50mm f/1.4, -1
70-200mm f/2.8L IS II, +2
100mm f/2.8L IS Macro, +8

So my two most used lenses the 50 and 70-200 were pretty much on the money, I've always been happy with them but it'll be interesting to see if they are just that little bit now, that may have been a small "problem I didn't know I had". The 24-70 I've mainly used for things where I didn't want too shallow DOF so it doesn't really suprise me I'd never noticed that one.

The 100mm macro was my last purchase and I hadn't been happy with autofocus, but being my first macro lens just put it down to the usual recommendations of tripod + manual focus for macro. After trying a few things like AI servo in case it was me moving too much I'd pretty much given up on handheld for anything approaching macro. After the adjustment I just wandered around the house doing a few hand-held test shots and now everything is bang on target.

Anyway that's a lens I see I'll get much more use from now, so I'm glad to have heard about the LensAlign and about focus adjustment in general which I must admit I didn't know about until I read this thread .

 

[Mt Spokane Photography]

After reading and reading again all your posts, I spent a couple of days to setup a good microadjusting session. I opted for Jeffrey Friedl's method, that was well explained and didn't require to wait (and pay...) for a lensAlign to be shipped. The chart is well designed and the logic of the light grey indicators seems quite reasonable.
The lack of an instrumental system for a perfect alignment camera target required some additional time to spend leveling and measuring, but at last everything was perfectly aligned.
I started with the 1Ds3 and the 100-400, followed by the 5D2 and the same lens. At the beginning everything seemed working well, following well every step of the procedure. Frequently tested if the sensor was reading just the right target, turned off IS, fast shooting time, manually scanned from minimum distance to infinity before AF, camera on heavy tripod. Both cameras-lens seemed affected by a slight front-focusing corrected at -5 and -6.
Then, just for checking, I tried to set at -9 with the unlikely result to get again a front-focusing. Tried and tried again, getting similar results.
After reading an old article by Tom Jackson about the disalignment between the real sensor and the focus outlines in the viewfinder, I convinced my self that a lab procedure for microadjustment is risky and too exposed to minimal error factors.
So I reset everything and tomorrow I will go for what seems to me a more empirical and real world approach.
I will puth cameras and lenses on my tripod, aiming at the right distance for a wall plenty of contrasting and fine details. Then I will take 41 pictures focusing only with the central spot, adjusting from -20 to +20 and I will compare the results in LR3.4.
Quite rude but nor too time consuming, and it seems to me the exact procedure for what the MA system was designed...
Any opinion?

I use Jeffreys chart as a starting point, and then verify with real world photos at various distances at full aperture. Sometimes a little tweaking is needed, but its narrowed down any errors by quite a bit, usually to +/- 2.

I do plan to buy a lens alighn unit, but i never seem to get around to it, and all of my lenses produce very sharp images.

 

[aldvan]

I received yesterday my LensAlign. Adding custom taxes and shipping the cost is unreasonable, per se, but the idea seems very good. I'm waiting for a relaxed moment of free time to MA my lenses...

 

[aldvan]

Today I started the MAF campaign. I have some doubts and hope that neuroanatomist will solve them for me.
Aiming at the two holes for alignment, particularly with shorter than long tele lenses, as a 100 macro at the required distance of 25*focal lenght, it is impossible, due to parallax, to have both of the two holes perfectly centered. The rear ones are both perfectly visible trough the front ones, but slightly eccentrics. Is it ok anyway? I can't see any method to have both of them perfectly centered at short distance.
The 100 Macro IS seems ok with +4 and a real world test seems confirm that.
Second doubt. I found different answers in order to compensate MAF. To put it in a simpler way, if the perfectly focused point is behind the zero on the ruler (backfocus?) the adjustment should be + or -? I tried many times with controversial results... So, with LensAlign the 100-400 seems backfocusing, but with a backfocus adjustment of +14 (ok for the LA), the backfocusing gets worse in the real world...
Thanks in advance...

 

[Loswr]

Today I started the MAF campaign. I have some doubts and hope that neuroanatomist will solve them for me.

I'll do my best...

Aiming at the two holes for alignment, particularly with shorter than long tele lenses, as a 100 macro at the required distance of 25*focal lenght, it is impossible, due to parallax, to have both of the two holes perfectly centered. The rear ones are both perfectly visible trough the front ones, but slightly eccentrics. Is it ok anyway? I can't see any method to have both of them perfectly centered at short distance.
The 100 Macro IS seems ok with +4 and a real world test seems confirm that.

I had to do a little digging, and it turns out that the LensAlign website claim that, "The graphics show the Pro model, however the MkII works exactly the same," is not quite true. I assume you have the MkII version, as that's what is currently shipping. I'd recommend aligning the hole on the left, i.e. the one in the center of the focus target, and let the other red circle be eccentric in the opening.

The Pro model has 5 holes vs. two, one in the center and two on either side of that, and I usually try to ensure that the red bullseye in the middle is centered, and the ones to the side are displaced symmetrically (something you obviously cannot do with only one other hole).

Second doubt. I found different answers in order to compensate MAF. To put it in a simpler way, if the perfectly focused point is behind the zero on the ruler (backfocus?) the adjustment should be + or -? I tried many times with controversial results... So, with LensAlign the 100-400 seems backfocusing, but with a backfocus adjustment of +14 (ok for the LA), the backfocusing gets worse in the real world...

If your lens is backfocusing, i.e. the plane of sharp focus is behind the 0-point on the ruler, then you need to apply negative adjustment to correct it, i.e. -6 or -14, depending on how the severity of the backfocus. An AFMA of +14 would make backfocus substantially worse.

 

[aldvan]

Thank you very much, neuroanatomist... After the first deluding attempts, I spent an hour to refine the procedure, and everything seems to go very well, now. Actually it needs a very longand accurate tune up, before starting, but after tha,t it is easier. The final result is 1Ds MkIII + 100-400 MAF +4, 5D MkII + 100-400 +2. Tomorrow I'll test this result in real world, and I'll tune up the other lenses...
During this kind of procedure you realize how flimsy are also good tripods. I have a Carbon Fiber and an heavy aluminum Manfrotto: I had to shoot with 10" timer to get a perfectly static image...

 

[PeterJ]

Aiming at the two holes for alignment, particularly with shorter than long tele lenses, as a 100 macro at the required distance of 25*focal lenght, it is impossible, due to parallax, to have both of the two holes perfectly centered.

I noticed the same when doing my 100mm macro, on the back of the Lensalign Mk II the outer alignment hole is marked macro target so I used that one. I wasn't 100% sure though so tried the center one as well but results were too close to call, so it doesn't seem to make much difference. I'm guessing the outer one is just marked macro in case you're so close the center hole and ruler aren't visible at the same time.

 

[aldvan]

Aiming at the two holes for alignment, particularly with shorter than long tele lenses, as a 100 macro at the required distance of 25*focal lenght, it is impossible, due to parallax, to have both of the two holes perfectly centered.

I noticed the same when doing my 100mm macro, on the back of the Lensalign Mk II the outer alignment hole is marked macro target so I used that one. I wasn't 100% sure though so tried the center one as well but results were too close to call, so it doesn't seem to make much difference. I'm guessing the outer one is just marked macro in case you're so close the center hole and ruler aren't visible at the same time.

I suspected that but I wasn't sure what he intended with 'macro target'...

 

[Viggo]

Neuroanatomist:

I took home with me a Spyder LensCal from work, and when using it, one actually doesn't need the distance I thought it would (comparing to the northernlight gif-image, pixel-allignment). And I adjusted 6 lenses on two bodies with half an hour, and they have all worked perfect since! Best tool you can buy ever.....

Thanks!

 

[Loswr]

Glad to help!

 

[aldvan]

Final results: ISO 800, max opening, 10" timer, max focal lenght, central spot

1Ds MkIII
16-35 f/2.8 L II 0
24-105 f/4 L +8
100 f/2.8 L Macro I +4
100-400 f/4-5.6 L +4

5D MkII
16-35 f/2.8 L II +3
24-105 f/4 L +2
100 f/2.8 L Macro I +6
100-400 f/4-5.6 L +2

It's evident that both the systems, withe very lens, are front focusing, and in a different way. So, I guess, the constant result should be charged mainly to the body instead to the lenses.
It's interesting to note that measurement was a pleasure with the 1Ds, showing a repeatable, constant and very precise progression from backfocusing to front focusing. The 5D was more unpredictable, showing a slightly random series of results, not always well connected between adjustment and focusing...