Help with lens calibration
I found this method for lens calibration
However, I'm confused, and need some help to make sure I'm doing this right.
I've never tried this method, and I've never seen it recommended by serious shooters. I have no idea how effective it is or isn't, but I'd start with suggesting a non-LCD target, fixed to a flat surface, and making sure your camera is perfectly parallel to the target, and that you're using an appropriate subject distance.
The appropriate subject distance is a source of debate, as you'll see anything from 20x the focal length to 40x the focal length to 6-10 feet to "the distance you most commonly use the lens at." The most important take-away would be don't use a distance near closest focus or infinity.
That said, if you only use a certain lens at a certain focusing distance, makes sense to test it somewhere near there. In general, shorter lenses are focused closer and longer lenses are likely to be focused further away. I might test a 17-35 at a few feet away, but I can't even focus the 70-200 that close. And if I tried to test 17mm at at 15 feet the DOF would be so significant that I wouldn't be able to tell if it missed focus!
Use horizontal lines/pattern for testing, in general. The cross-hatched sensors can pick up either orientation, but the vertical sensors can't pick up vertical lines very well, so there's no sense in using them for a target.
I've tested two lenses, nikon 17-35 f2.8, and tamron 24-70 f2.8 vc, both at 26mm
Be aware that zoom lenses will focus different at different focal lengths. Each lens is different. Don't think in terms of 26mm vs 26mm. Think of each lens on its own.
As far as I understand, I setup the pattern on my lcd screen, set up the camera at the proper distance, go into live view and with the lens in AF, allow it to focus on the pattern. I try this several times over, and the meter on the lens shows the exact same focal distance every time, dead on.
The way to check focus is not by looking at the lens(!!!!), but rather examining the images on-screen. (Yet another reason not to tie up your computer screen with the testing.)
Then, I switch out of live view, and hit focus. The lens is not consistent every time. The focal distance jumps around and rarely seems to hit the same focal distance twice in a row. Granted it is very small discrepancies but it seems it's enough to keep AF from getting a sharp focus consistently. It did the same on both lenses.
The correct way to test focus is to rack the lens out of focus--infinity is usually recommended--and let it refocus. Take a picture. Compare that to the live-view-focused image. Or, if you're using a ruler-equipped target, look for front or back focus. And you're going to have to take several images to get an average--you can't rely on just one image. If you want to take it one more step, shoot several images from both directions--meaning shoot several with the lens racked to infinity and several with the lens racked to closest-focus.
So, my first question is, is this normal? Does the live view use something different to determine focus than non-live view?
Yes. Live view is contrast-detection AF. Done properly it is more accurate that PDAF (phase detection). That is why it is often used as a reference to compare to PDAF. (Though clearly live view has limited application for many shooting scenarios.) The disadvantage to using it is that it only gives you a comparison, not a direction. IOW, it doesn't show you front or back focus, only that the PDAF shot and the CDAF (live view) shots look different. This is one advantage to having a ruler or some sort of depth gauge in your test, as it can tip you off a little quicker as to which way the lens is trending and how severely. The ruler method isn't perfect either, but it has some pluses that way.
Second question, although it was hard to tell because the 17-35 lens didn't seem to be consistent, it seemed to be consistently front focusing when compared to live view. I pushed AF fine tune all the way to +20 and still seemed to be front focusing.
Welll...here's the thing: if you're doing the test right, it will be fairly consistent across an average of several frames. So if it isn't consistent I would start with looking at the your testing methodology. I'm not saying it will be exactly the same every frame, but it shouldn't be wildly different between frames either. If it is, you may very well have a mechanical problem (lens and/or body). But start with examining your testing, as that will benefit every lens you test, not just this one.
Third, does AF fine tune ONLY effect the AF of the lens when NOT in live view? It seemed to not change a thing when focusing IN live view.
Correct. Fine-tune only affects PDAF, not contrast-based AF.
AM I DOING THIS RIGHT?
I recently picked up the 17-35 and my latest photo, once I got into lightroom with the raw files, seemed soft, so I was compelled to test.
You might very well have an issue, but you're going to have to do a lot more very careful testing to narrow it down. I've never yet had a nikkor (out of maybe a 20 in the last decade) that didn't fall within the fine-tune range. Some have been close (>15), but still correctable. That isn't to say it can't happen, and maybe I've been lucky.
Fine-tune is a blessing and a curse. It's a bit of a pandora's box in that once you start you find out just how many variables can be impacted--focal length setting on a zoom, whether or not you're near the close-focus or the infinity end, parallel-ness to the target, the kind of AF target that you're using, etc.
There's never going to be one perfect setting that makes everything work perfectly all the time. If a lens needs different correction at one end of the zoom range compared to the other, than what to do? Split the difference? But what if you use one end far more than the other?
These are the sorts of things you'll have to work through and come up with what works best for you. And this is precisely why one common advice is "turn it off and forget about it." Because if you don't do it exactly right, you have the possibility of making the focus worse in some cases rather than better. Let's say one end of a zoom likes -2 and the other end likes +4. Well if you only cue in to one end or the other....either setting makes the other end a +/-6, whereas leaving it 0 made the tuning never off by more than +/- 4, and in many scenarios less than that.
You said you're seeing issues with your 17-35, so I think it's worth exploring. But be patient, work on the testing methodology, and realize that this may take many hours of work to completely iron out. And, if you're going to do it, you have to go all the way, or else you run the risk of setting the fine tune incorrectly and actually making the AF worse.
I LOVE fine tune, but it's not always easy, it's tedious, and it's not for everyone.
|Post (hide subjects)||Posted by||When|
|Jan 6, 2013|
|Jan 6, 2013|
|Jan 6, 2013|
|Jan 6, 2013|
|Jan 7, 2013|