|Home||About us||Commercial Photography||Print Gallery||Articles/reviews/blog|
Specialist focus adjustment and accuracy check
We've been having a look at the new SpyderLensCal from Datacolor - it's a tool that allows you to measure aspects of the autofocus accuracy of your camera.
Keith has had a go at testing the SpyderLensCal, using his Canon EOS 1Ds mark 3 camera and various autofocus lenses.
As someone who has only relatively recently started using autofocus cameras, I'm still slightly impressed by the fact that AF works at all.
That said, my first 'pro' camera was a Canon EOS 1Ds, so perhaps I'm a bit spoilt when I read all the complaints and discussions of AF performance on various forums.
Any AF camera system has to contend with errors and tolerances in the various components that are part of it.
Lenses vary and camera bodies vary - so the final 'correctly focused' decision of the AF system may be slightly in error.
With my Canon 1Ds, the only way of getting everything fully optimised was to send your camera and lenses off for calibration.
Indeed, if you find that every single lens mis-focuses by a similar amount, it's a good sign that your camera might need servicing.
During the years I used my 1Ds I never noticed any focus errors that could not more readily be attributed to user error.
However, when I first got the 1Ds mark 3, I tested all my AF lenses and found that I could fine tune some of them a bit to improve their focus accuracy on that camera body.
I've written up some of my experiences with this process in another article about Camera AF Microadjustment.
I'll come back to some of my thoughts about the various different methodologies in the conclusions.
The SpyderLensCal comes in flattened form in a retail pack.
It costs $59 / 49 euro
The three parts, are:
There is a 'quick start guide' giving an outline of the steps you take to perform the adjustment, but no software or any thing else to guide you.
There are additional resources on the Datacolor web site, but you are going to need to find out how to set individual AF adjustments on your own camera.
Yes, this does mean reading the manual...
If you are someone who doesn't read manuals and expects things to 'just work' then I'd seriously suggest you get assistance if you want to try this.
Let me say again, that microadjustment is completely irrelevant to a large proportion of people who take photos - I'm assuming that if you are reading this then you are comfortable with the details of your camera's operations.
I'm going to show an example testing a Canon EF24-70 2.8L lens.
It just so happens that I've no idea of how good this lens is, since it belongs to CPS and is on loan whilst my normal 24-70 is away being repaired (a long saga that I'll leave for another day...)
The next photo (right) shows how the target hinges up and is locked perpendicular to the base, by the ruler.
The ruler was a bit stiff to release when I first took the device out of the box.
On the underside of the ruler is a ridge that holds things in place.
The line from the number '0' should line up with the front of the target.
There is a Video about the product at the Datacolor web site.
You need a camera that supports AF adjustment.
Currently (2010), these models offer it:
The device is attached to the top of a tripod, and levelled using the bubble level.
It's important to get things lined up carefully.
You need to be square on to the target.
I've set the camera up about 3.5 metres from the target, on a second tripod (below - move mouse over image).
In Canon's own advice about AF adjustment, they suggest putting the target at least 50 times the focal length of the lens away from the camera.
I'm doing the adjustment at the longest focal length of the lens - 70mm, so the distance should be at least 3500mm (50x70mm)
Here's the camera and lens I'm using, with a cable release to minimise movement.
Before I take each test shot I move the lens focus manually to infinity, just to start each test from the same point.
You might wonder if I'm using things like mirror lockup to further reduce vibration and improve image sharpness?
I'll come back to just how much accuracy is really needed in the conclusions, but I'll just say that AF microadjustment is not akin to setting up your own optical test laboratory.
When setting things up, I make sure that the centre AF point is selected and that it is focusing on the centre of the target.
The somewhat brief instructions suggest taking photos and zooming in with the rear screen to view the ruler and see whether the front or rear section is sharper.
If the front is sharper than the target, then the lens is 'front focusing' and you can use the camera's adjustment procedure to alter it.
Here's the view after I've taken a photo.
It's actually a bit easier to see that the lens is slightly back focusing in this picture, than it was when looking at the rear LCD.
After some experiment, I decided to just take a series of photos at microadjust (MA) settings of +20, +15, +10, +5, 0, -5, -10, -15 and -20 and see what they looked like.
After taking the shots, I processed them in Adobe Camera Raw (with no sharpening or noise reduction) and cropped out just the target.
The images below are the output from the conversions, at 100%
As you can see, sharpness at full aperture (f/2.8) isn't great (I use the 24-70 typically at f/7 or thereabouts, and almost always have some sharpening applied at the RAW processing stage)
Oh dear ... is this perhaps not as clear cut as you might have hoped?
These are real images at 100%, from a real test at a good testing distance.
By using a quick preview of the images. I decided that -5 was the best and that 0 looked better than -10, not by much, so I picked a value of -3, which looks just fine.
It's difficult to show here, so I've animated a sharpened version of the images, stepping from +20 to -20 in increments of 5.
I think it's this imprecision that comes to a surprise to people expecting to see some dramatic improvement.
You might try applying a small amount of capture sharpening to each image and see if it helps show up differences.
Fortunately with a RAW converter such as ACR I was able to consistently apply different conversion settings (same to each image) and see whether it made it easier to see a difference.
Remember that you are processing these image to show detail in the target, not take a nice photo of the room behind it.
If I wanted to be more precise, I'd probably do a series from 0 to -10 in steps of 2.
When I originally tested my own 24-70, it needed no adjustment at all.
Microadjustment definitely works if your lenses need it, however at the back of my mind, I do wonder just how many people really do work that requires such precision set-up.
The only times I'm really picky about detailed focus is in my architectural and detailed product photography, however I'm mostly using manual focus lenses for this, so AF is pretty irrelevant.
My longest AF lens is a Canon EF70-200 2.8L IS and it is virtually much spot on in its AF. If I was using an 800mm with a 1.4x converter on a 1D Mk4 taking photos of birds eyes, I might pay a bit more attention...
Some time ago I wrote my original article about autofocus microadjustment, and included a screen test pattern that you can use for making adjustments. As it happens, this worked well, when displayed on my Apple 23" cinema display.
In the time since, I've had numerous mails thanking me for the information in the article and saying how it helped people make their adjustments. However, I've seen a number of comments on forums, from people who've just not got it to work.
During the testing of the SpyderLensCal I put up the test image on my new 15" MacBook, with its high res display.
At the same distance as the test target, it was very difficult to see a good sharp pattern. If you add to this the broader depth of field of an f5.6 'kit lens' at its longest focal length, I can see why some people have had difficulties.
So, in the test setup, to the right, with the target at ~50 times the focal length, the test pattern on the computer was less precise.
As someone who's championed the test pattern approach, this came as a bit of a surprise. Going back to my big Apple display however, it worked somewhat better. It also worked better on my old 15" G4 Mac PowerBook with its lower resolution screen.
The difficulty would appear to be partly due to the relatively small target size in the image when at 50 times the focal length.
The SpyderLensCal does indeed show a more visible change in focus when I reduced the distance to just over 2 metres, but I'm reminded that Canon probably suggests FL x50 for good reasons, and I wouldn't want to test things too much closer.
If you look at the camera/target distance in the Datacolor video, it's just too close - this is why the AF adjustment is so obvious. There are no suggestions as to the correct distance in the quick start guide.
If you're testing long lenses, with teleconverters, then it's more likely that you'll do it outside, where the easy to set up (and waterproof) SpyderLensCal will benefit from brighter lighting (better contrast).
The device is capable of sitting on a flat surface, if you don't happen to have a second tripod available.
This note about the adjustment process is from Canon who suggest taking three shots for each setting:
Thank you for supporting our site.
I mentioned earlier that you should be careful in setting up the SpyderLensCal to be level, and to set your camera up so that the target is in the centre of the frame.
The way the device is constructed, offers several important visual clues you can use to be happy with the alignment, such as the edge of the ruler against the edge of the target.
But how precise do you -really- need to get?
Much as with aspects of colour management, there is an element of making precise adjustments and calibrations that appeals to a certain group of photographers (invariably male ;-).
When approaching the setup of camera and target in a similar manner to setting up an optical test lab for a lens manufacturer, the sloppy imprecision of the plastic SpyderLensCal is just not good enough.
If only it were made of precision metal parts and cost several times as much, then it -must- be better...
After trying out the SpyderLensCal, and seeing just how vague the difference in a few steps of adjustment settings were, I have to say I couldn't see any benefit if it was made from stainless steel and had built in laser alignment devices.
I'm doing a basic lens adjustment on my camera, not coming up with a foolproof method of testing mirrors for space telescopes.
The SpyderLensCal did impress me with its design simplicity, and I'll now include it with my other methods of checking autofocus accuracy.
If there was one group I'd really recommend the device to, it would be photography clubs.
If you've just one lens it might not be worth your while. At a club it would be a great resource, not to mention the opportunity for people to learn a bit more about what their cameras can do (and can not do ;-)
Costs $59, or 49 Euros (excl. VAT)
Buying anything from Amazon (not just what's listed) via any of the links below helps Keith and Karen keep the site going - thanks if you do!
Add your comments and questions
Real names get answers... Please note that I like to know who I'm talking to (just like in real life) so anonymous comments are rather more likely to be ignored or summarily deleted ;-) Note that comments with links may also get held upblog comments powered by Disqus
The views in this article represent those of Keith Cooper.
Keith is always happy to discuss matters raised in his articles. You can Email Us
Northlight Images prides itself on its independence when giving advice. We do not sell hardware or software and have no direct commercial links with any of the software or hardware vendors that may be mentioned here. See our Review Policy for more information.
Our reviews never recommend the 'best product' for you - see why this matters at:
You can search all the many hundreds of articles and reviews on the site for more information
Have you found an article on the site useful or helpful?
If so, please consider sharing a link to the article or mentioning it on a forum or blog - Thanks to everyone who's helped the site become better known.
Explore our site... Digital Black and White photography and printing - some of Keiths thoughts, techniques and tips for those interested in a digital approach to black and white. There are many hundreds of entirely free articles and reviews on the site. New site content appears on the News, articles and reviews page.
Northlight Images is based in Leicester in the UK and supplies Commercial Photography services.
Visiting Leicester? - see hundreds of Keith Cooper's local photos at our Leicester Photos site
Commercial Photography - UK Architectural Photography services - Industrial - Print Gallery - Landscape Photographer