Luminous Landscape Forum
Equipment & Techniques => Digital Cameras & Shooting Techniques => Topic started by: xpatUSA on September 05, 2013, 09:20:49 pm
-
Hello,
I've read about front/back focus problems and, having recently acquired a used DSLR, I decided to test it and my favorite lens with the slanted ruler method.
I've edited the subsequent content out of this original post due to the peculiar response below. The content was apparently too technical [for kaelaria, see below] and not "real world" enough. Perhaps a moderator would kindly delete this whole topic.
Sorry for any inconvenience to more normal members. The original post showed a ruler shot and basically asked if it was acceptable but I showed too much information about ruler angle, divisions, and more.
Since then, I've found that test results can vary with the direction of the wind, can be inconsistent, etc. And that some folks care about front/back focus and some do not. And that the scene's subject matter matter.
The reason for the test was that a camera I bought seemed to be a bit soft - this from a camera with no AA filter and with both a kit zoom and with a good macro prime on it.
-
If you never noticed it before doing the test - then it's perfectly fine. If you only think you have a problem because you are data peeping then just stop ding that. If you saw focus issues before hand, how does the data correlate to what you were experiencing? Too many people get wrapped up in numbers with no real world relevance.
-
If you never noticed it before doing the test - then it's perfectly fine. If you only think you have a problem because you are data peeping then just stop ding that. If you saw focus issues before hand, how does the data correlate to what you were experiencing? Too many people get wrapped up in numbers with no real world relevance.
I live in hope that this aggressive response is not typical of this forum.
-
Truth is not aggressive.
-
Truth is not aggressive.
How cute.
-
If you never noticed it before doing the test - then it's perfectly fine. If you only think you have a problem because you are data peeping then just stop ding that. If you saw focus issues before hand, how does the data correlate to what you were experiencing? Too many people get wrapped up in numbers with no real world relevance.
and how are you supposed to know if you have a problem if you don't peep? I'm not sure what the original post was, but the fact is almost every dSLR has a back or front focus issue, there are just so many variables and tolerances are tight, and autofocus has a pretty large "tolerance" which often isn't that great. A photographer who fails to recognize this and calibrate their equipment may be very disappointed when they suddenly find out that awesome image they want to print large won't hold up or when then need to do some aggressive cropping.
We calibrate cameras almost every day, and only about 5% are in focus with no adjustment. You can only tell it if you "peep". To me what some label "peeping" is just standard practice in a quality workflow. In my workflow, every image at some point will be examined at 100% early to insure it will hold up to printing and often later as I fine tune creative sharpening.
I understand the entire discussion about technical vs creative, but mastering the technical is an important part of not letting it get in your way when creating. I agree too many get wrapped up in "numbers" (DxO comes to mind), but to me fine tuning your gear is important, and peeping is how you do that.
-
I live in hope that this aggressive response is not typical of this forum.
If you think that is aggressive then you are in for a shock. If that is the worst that has happened to you then you are lucky. :D
-
I get the impression this issue is a minefield. One carefully calibrates the autofocussing by shooting a nearby object like a slanted ruler, only to find that focussing at infinity is now worse. And this is for a fixed focal length lens.
With zooms, one may not only get a variance of front and back focussing depending on focussing distance, but also depending on the focal length setting.
Add to this the fact that autofocussing precision, whether or not the AF Fine Tuning has been used, is rarely exact, then that old advice of "F8 and be there" still holds.
-
I get the impression this issue is a minefield.
That's true to an extent. However automated programs like Focal Pro make investigation of the issues involved a pretty straight forward, once you know what all the actual errors are you can decide what, if any, changes to make.
Whilst it can be subtle, it can also make the difference between 'just about acceptable' and 'good'. I'm happy I've done the work to make focusing 'good' more often.
"F8 and be there"
Fine if you want everything at f8 and have enough light for it.
The other side of testing is that you might find that f8 isn't the best stop to use anyway. A lot of my lenses perform better at wider apertures than f8, testing has ensured that I nail the focus more reliably at those wider apertures too.
Don't knock technologies like AFMA until you've exploited them. It can all be part of the small differences that separate the ordinary from the excellent.
-
Don't knock technologies like AFMA until you've exploited them. It can all be part of the small differences that separate the ordinary from the excellent.
Hi,
I agree. Not only do my lenses improve their focus acuuracy over the entire focus range, but I also get the impression that focusing is faster. One typically calibrates for a commonly used distance anyway, so even if the lens requires different settings for different distances, chance has it that the adjustments are not too different from the most common case.
Zoom lenses may require a compromise, but even there one can usually find an average adjustment that improves focusing across the range. Even with lenses that exhibit focus shift when changing the aperture, one can often find an improvement around the more commonly used aperture setting, an adjustment which will differ from the wide open AF optimum.
Cheers,
Bart
-
and how are you supposed to know if you have a problem if you don't peep?
What a strange point of view. (in a very strange thread)
To my way of thinking, photography is a visual art - so any "fault" that cannot be visually discerned in the normal output is, from any practical point of view, totally irrelevant.
Now, of course, the key words in that statement are "normal output" and that may be different for different photographers. The output of a landscape photographer is different from that of a portrait photographer or a wildlife photographer or a product photographer or a macro photographer...etc., etc. So a fault that may matter to one may be completely inconsequential to another.
But to go looking beyond your own normal output for faults, real or imaginary, is a pretty pointless exercise.
So, I don't regard the second post on this thread as in any way "aggressive" but, rather, as a perfectly reasonable response to the problem (or non-problem) of the OP.
-
any "fault" that cannot be visually discerned in the normal output is, from any practical point of view, totally irrelevant
What's 'normal output' ?
I think a lot of people can be aware something is wrong with an image without quite knowing what the problem is. Looking closely often reveals the problem, if you can then fix it and things improve, great.
If pictures aren't as sharp as you'd like, investigate.
Was the best focus where you'd expected ? If not, was it user error or technical fault ? If it was a technical fault can I improve it ?
Would it be sharper if you'd used a different aperture ?
Is the lens just not very good ?
There's nothing wrong with trying to get the best out of kit, it would be crazy not to.
Some people do obsess about it, but most of us test once and get on with shooting the best pictures our kit will allow. If you're happy with sub-optimal results, that's up to you, but don't complain when some people strive for excellence.
-
What a strange point of view. (in a very strange thread)
To my way of thinking, photography is a visual art - so any "fault" that cannot be visually discerned in the normal output is, from any practical point of view, totally irrelevant.
Now, of course, the key words in that statement are "normal output" and that may be different for different photographers. The output of a landscape photographer is different from that of a portrait photographer or a wildlife photographer or a product photographer or a macro photographer...etc., etc. So a fault that may matter to one may be completely inconsequential to another.
But to go looking beyond your own normal output for faults, real or imaginary, is a pretty pointless exercise.
So, I don't regard the second post on this thread as in any way "aggressive" but, rather, as a perfectly reasonable response to the problem (or non-problem) of the OP.
I guess we disagree, because I think your response is odd . Making sure your equipment is performing at it's optimum before using it is pretty standard practice for anyone seeking maximum quality. Many of us test a lens before we buy it, or shoot with a system before we invest in it - a common practice for decades (nothing to do with digital)l. I completely agree that many get way too wrapped up in the technical, but this issue doesn't really belong in that discussion.
Very few photographers, especially those who enjoy being creative and shooting all types of photography, have a predetermined "normal". I hear this all the time, and sorry, just don't buy it. It's the old "if your really an artist you can't let the technical get in your way". I sit here watching customers come in all day (some of them pros and semi pros) asking me to get more out of an image than I can because they didn't use a tripod or used a poor lens - or it is just not quite as sharp as it could have been. It's so easy to calibrate your system, why wouldn't anyone not want to do it? Once you do it, your done till you buy a new body or lens.
I get the impression this issue is a minefield. One carefully calibrates the autofocussing by shooting a nearby object like a slanted ruler, only to find that focussing at infinity is now worse. And this is for a fixed focal length lens.
With zooms, one may not only get a variance of front and back focussing depending on focussing distance, but also depending on the focal length setting.
Add to this the fact that autofocussing precision, whether or not the AF Fine Tuning has been used, is rarely exact, then that old advice of "F8 and be there" still holds.
logic is sound, and yes even this isn't perfect but what you are compensating for the the autofocus precision issue you mentioned ... the relationship of what the focusing sensors think is sharp vs what is sharp on the sensor itself. No matter how tight you try and manufacture the tolerances, this is just asking too much to get the distances of the two paths exactly the same, and to be out of focus doesn't take much. So the offset compensates for this issue on a camera by camera/lens by lens basis.
Regarding zooms, yes still an issue, although if you calibrate a zoom you will see improvements in the entire range. the 5D 3 now has the ability to correct both ends of the zoom range. For those needing perfect focus, live view is still by far the best option.
-
I should have written can be a minefield. One little word makes a difference. ;D
I'm not trying to imply that AF fine-tune is not a useful feature to have. Of course it is. There are some lenses on some bodies that are just a little bit out and a simple AF fine-tune adjustment results in a worthwhile improvement without complication.
Before the feature of AF fine-tune was provided on DSLRs, I would simply return a lens if it didn't autofocus accurately. Perhaps my frustration in this regard at the moment is due to my not having returned a recently bought lens as soon as I discovered it wasn't focussing properly on my D7100. I thought I'd fixed the problem with a fine-tune adjustment of +10, using a fairly close target, but later discovered that this fine prime lens, the Sigma Art 35/1.4, was producing abysmal results at infinity, on the D7100.
It required a fine-tune adjustment of the maximum of +20 for distant subjects. The difference between the +10 adjustment and the +20 adjustment at infinity is very obvious, particularly between the apertures of F1.4 to F4.
Now, the whole purpose of buying a lens like the Sigma Art is because it's ultra sharp and can produce a shallow DoF at F1.4. I can see no point in using a compromise AF fine-tune setting of +15, unless one is prepared to use F8.
When I use the lens for close-up work, I'll have to remember to check that the AF fine-tune is set on +10. When I use the lens for landscape, I'll have to check that it's set to +20, or use a tripod and LiveView. However, if I don't want to bother fussing around making changes to such adjustments according to the circumstances, I could just leave the setting on +15 and use F8 all the time.
Another complication I've noticed is the inherent inaccuracy of the autofocussing system regardless of AF fine-tune considerations. I use the AF-ON button and a single focussing square so that focussing is independent of metering. By pressing again the focussing button each time one takes a shot of the same scene, from the same position, focussing on the same spot, one gets an indication of the focussing inaccuracy of the system. It's as though one has altered the fine-tune setting for each shot. However, such differences are only apparent at wide apertures. At F8 there's no problem.
-
Perhaps my frustration in this regard at the moment is due to my not having returned a recently bought lens as soon as I discovered it wasn't focussing properly on my D7100.
Yes, from your description you should have just returned it as faulty.
This is one advantage of being able to test and quantify problems, retailers won't argue if you can prove faults with kit. You should never have to live with kit that not fit for purpose.
-
Measuring something that can otherwise _never_ be detected in "real-world" use may be counter-productive.
Measuring something that can otherwise only be detected once every 2 years, when you decide to print that special image big, sounds like a sensible thing to do.
If you do not know beforehand how big you are going to print (or how much cropping, shadow push etc), then it seems reasonable to do the appropriate measurements (or read about other peoples measurements) and apply some margin. The upside is that your images may have some (potentially very small) improvement in quality. The downside is that you have to spend time and resources that could otherwise be spent taking images or with your family. Choose accordingly.
-h
-
Hi,
Than you get a new camera that is propely adjusted, and now you realize that all of your images of yore are subpar...
Best regards
Erik
What a strange point of view. (in a very strange thread)
To my way of thinking, photography is a visual art - so any "fault" that cannot be visually discerned in the normal output is, from any practical point of view, totally irrelevant.
Now, of course, the key words in that statement are "normal output" and that may be different for different photographers. The output of a landscape photographer is different from that of a portrait photographer or a wildlife photographer or a product photographer or a macro photographer...etc., etc. So a fault that may matter to one may be completely inconsequential to another.
But to go looking beyond your own normal output for faults, real or imaginary, is a pretty pointless exercise.
So, I don't regard the second post on this thread as in any way "aggressive" but, rather, as a perfectly reasonable response to the problem (or non-problem) of the OP.
-
Yes, from your description you should have just returned it as faulty.
This is one advantage of being able to test and quantify problems, retailers won't argue if you can prove faults with kit. You should never have to live with kit that not fit for purpose.
I left it too late. I assumed I could fix any focusing problems with AF fine-tune, and in a sense I can, provided I take the time to readjust the fine-tune setting according to the conditions when using the D7100. At least the maximum setting of +20 on the D7100 is sufficient to get a sharp image at infinity.
Another issue which muddies the waters, is determining whether it's actually the camera body that's at fault, rather than the lens. This Sigma lens doesn't require any AF fine-tune on my D800E, although there are still issues with the lens on the D800E, such as uneven sharpness in the four corners at wide apertures.
However, the one solid improvement of this Sigma prime compared with my only other Nikkor lens of the same focal length (the 24-120 zoom set at 35mm), is the edge and corner sharpness at apertures between F5.6 and F11. It really is noticeably sharper in the corners than the zoom is at those f/stops, on the D800E. That's worth something.
-
Hello all,
I'm the perp of the original post.
Since this thread seems to be ongoing, I edited the OP again and put some explanation so as to make the thread a little less "strange".
Again, apologies for any confusion . . .
P.S. I've just noticed at bottom left of the page "remove topic" which I could have used when my contretemps with kaelaria got personal. Can Original Posters really wipe out an entire thread?
-
The downside is that you have to spend time and resources that could otherwise be spent taking images or with your family.
this takes very little time ... maybe 20 minutes? even a complex system like a tech camera which requires manually calibrating each lens to the body/back combination takes about 15 minutes per lens. This is once for each camera lens combination... certainly a pretty insignificant amount of time to invest in quality for something you will most likely use for quite some time (often years). I don't think it will infringe on much shooting or family time.
And if you don't want to take the time, many good camera stores offer it as a service. At my store, we will do it for you for $45 for the first lens, $25 for each lens done at the same time. Just drop it off and pick it all up the next day.
-
Just to chime in, I am finding that what works best for me (after using gizmos like the LensAlign, and software-driven solutions for a while), is to shoot that has good range of depth stuff wide open at the distance I shoot at, and fiddle with the AF adujustment so the focus is falling in the plane I'm focusing on. With the D800E I can do the image review right on the LCD, and it gets dialed in pretty quickly.
-
I should have written can be a minefield. One little word makes a difference. ;D
I'm not trying to imply that AF fine-tune is not a useful feature to have. Of course it is. There are some lenses on some bodies that are just a little bit out and a simple AF fine-tune adjustment results in a worthwhile improvement without complication.
Before the feature of AF fine-tune was provided on DSLRs, I would simply return a lens if it didn't autofocus accurately. Perhaps my frustration in this regard at the moment is due to my not having returned a recently bought lens as soon as I discovered it wasn't focussing properly on my D7100. I thought I'd fixed the problem with a fine-tune adjustment of +10, using a fairly close target, but later discovered that this fine prime lens, the Sigma Art 35/1.4, was producing abysmal results at infinity, on the D7100.
It required a fine-tune adjustment of the maximum of +20 for distant subjects. The difference between the +10 adjustment and the +20 adjustment at infinity is very obvious, particularly between the apertures of F1.4 to F4.
Now, the whole purpose of buying a lens like the Sigma Art is because it's ultra sharp and can produce a shallow DoF at F1.4. I can see no point in using a compromise AF fine-tune setting of +15, unless one is prepared to use F8.
When I use the lens for close-up work, I'll have to remember to check that the AF fine-tune is set on +10. When I use the lens for landscape, I'll have to check that it's set to +20, or use a tripod and LiveView. However, if I don't want to bother fussing around making changes to such adjustments according to the circumstances, I could just leave the setting on +15 and use F8 all the time.
Another complication I've noticed is the inherent inaccuracy of the autofocussing system regardless of AF fine-tune considerations. I use the AF-ON button and a single focussing square so that focussing is independent of metering. By pressing again the focussing button each time one takes a shot of the same scene, from the same position, focussing on the same spot, one gets an indication of the focussing inaccuracy of the system. It's as though one has altered the fine-tune setting for each shot. However, such differences are only apparent at wide apertures. At F8 there's no problem.
That is exactly the issue I had with mine which I returned immediately. It's a shame because I really want the lens. I will probably get another when work slows down.
I used af fine tune at a few feet to pick a setting. On landscapes I found infinity OOF even stopped down. The lens was great at close distances. If I am lucky it was a bad batch with a poor jig that has been fixed.
-
That is exactly the issue I had with mine which I returned immediately. It's a shame because I really want the lens. I will probably get another when work slows down.
I used af fine tune at a few feet to pick a setting. On landscapes I found infinity OOF even stopped down. The lens was great at close distances. If I am lucky it was a bad batch with a poor jig that has been fixed.
I sort of wish I'd never bought the lens because I've spent so much time stuffing around trying to get the AF right. Nevertheless, I have got a sort of workable solution. Re-adjusting AF fine-tune on the menu is not particularly troublesome, although one might occasionally miss a good shot whilst doing so. I guess this lens is only going to be useful for certain specialized, preplanned, situations.
-
Hello,
I've read about front/back focus problems and, having recently acquired a used DSLR, I decided to test it and my favorite lens with the slanted ruler method.
If your camera has live view, you can use the method (http://www.clarkvision.com/articles/microadjustment/) described by Roger Clark to check focus. Auto focus in live view is more accurate since it uses the image on the plane of the sensor as Roger describes.
Bill
-
If your camera has live view, you can use the method (http://www.clarkvision.com/articles/microadjustment/) described by Roger Clark to check focus. Auto focus in live view is more accurate since it uses the image on the plane of the sensor as Roger describes.
Hi Bill,
Roger's method works best with long focal lengths, and for lenses with modest widest apertures. Lenses with an aperture wider than f/2.8 may suffer from flatness of field issues, and possibly focus shift when stopping down. It's hard to predict, but it wouldn't surprise me if e.g. Ray's lens exhibits some of that behavior.
The AF aperture is limited at approx. f/2.8, so wider apertures will not contribute to the AF solution, but they may influence the overall optical focus result that the sensor records. Since AF is determined wide open, it may differ from the stopped down focus plane.
Cheers,
Bart
-
Hi Bill,
Since AF is determined wide open, it may differ from the stopped down focus plane.
Cheers,
Bart
Was feeling good about my Panasonic m4/3 until I read that, Bart. Although in manual focus I get 10X live view momentarily, the f-number number, as we know, reverts for the shot to whatever has been selected in the cam. Sadly, in my cam, pushing the button to preview "DOF" cancels the 10X view.
So, with two better quality Panasonic zooms and their Leica 45mm macro, I'm hoping to not notice front/back focus effects!
-
Was feeling good about my Panasonic m4/3 until I read that, Bart. Although in manual focus I get 10X live view momentarily, the f-number number, as we know, reverts for the shot to whatever has been selected in the cam. Sadly, in my cam, pushing the button to preview "DOF" cancels the 10X view.
So, with two better quality Panasonic zooms and their Leica 45mm macro, I'm hoping not notice front/back focus effects!
you shall not, speaking about 20/1.7, 25/1.4, 45/1.8, 45/2.8, 60/2.8, 75/1.8, 35-100/2.8 AF m43 lenses...
-
I sort of wish I'd never bought the lens because I've spent so much time stuffing around trying to get the AF right. Nevertheless, I have got a sort of workable solution. Re-adjusting AF fine-tune on the menu is not particularly troublesome, although one might occasionally miss a good shot whilst doing so. I guess this lens is only going to be useful for certain specialized, preplanned, situations.
Send Sigma an email explaining the problem, maybe they will take it under warranty service.
-
Send Sigma an email explaining the problem, maybe they will take it under warranty service.
Yes, I could do that. However, I've already had an unsatisfactory experience doing something similar with Canon equipment a few years ago, so I'm a bit reluctant to risk wasting more of my time.
I was very impressed with the sharpness of my Canon EF-S 17-55/F2.8 zoom, until I discovered that it wouldn't accurately autofocus at close distances when set at F2.8 on my new Canon 40D. I sent both lens and body to Canon service under warranty, to get the problem fixed. They were unable to fix it and claimed that both body and lens were within specification.
The Canon 40D didn't have the AF fine-tune feature, so I was stuck. But within a short period, Canon released an upgrade which did boast the AF fine-tune feature, the 15mp 50D. So I bought it, not just for the AF fine-tune, but that was a major attraction.
The irony is, the 50D focused perfectly at maximum aperture with the EF-S 17-55/2.8, without any AF fine-tune adjustment at all. I can only assume that the AF fine-tune has a dual purpose. To compensate for any slight maladjustment of the lens, and/or to compensate for any slight maladjustment of the camera body.
Both lens and body could be within specifications, that is, within the acceptable margin of error, but there could be situations where the errors in both body and lens are at the same end of the allowable spectrum of error, and have an additive effect.
Conversely, there could be other situations where the errors in both the lens and the body are at opposite ends of the spectrum and cancel each other out, creating the false impression that one is lucky to have an ideal camera body and lens without error.
-
Both lens and body could be within specifications, that is, within the acceptable margin of error, but there could be situations where the errors in both body and lens are at the same end of the allowable spectrum of error, and have an additive effect.
Conversely, there could be other situations where the errors in both the lens and the body are at opposite ends of the spectrum and cancel each other out, creating the false impression that one is lucky to have an ideal camera body and lens without error.
That's my understanding too.
-
Hi Bill,
Roger's method works best with long focal lengths, and for lenses with modest widest apertures. Lenses with an aperture wider than f/2.8 may suffer from flatness of field issues, and possibly focus shift when stopping down. It's hard to predict, but it wouldn't surprise me if e.g. Ray's lens exhibits some of that behavior.
The AF aperture is limited at approx. f/2.8, so wider apertures will not contribute to the AF solution, but they may influence the overall optical focus result that the sensor records. Since AF is determined wide open, it may differ from the stopped down focus plane.
Cheers,
Bart
Bart,
Your post brings up another problem with AF fine focus adjustment: many large aperture lenses exhibit focus shift when stopped down due to spherical aberration. This can be pronounced with f/1.2 lenses. The peripheral rays (those farthest from the optical axis) intersect closer to the lens along the optical axis than those closer to the optical axis, and the focus shifts further from the lens as the lens is stopped down. I'm not sure what you mean when you say AF is determined wide open. DigLloyd has a good section on focus shift in his Making Sharp Images (a pay site, but well worth the modest cost). He says that Canon AF sees the image effectively at f/5.6 and does not utilize the peripheral rays. He says Nikon is somewhat better in this regard, but does not specify by how much.
The result of these considerations is that if one uses AF to focus with the Canon on a subject at infinity using the 50 mm f/1.2 lens, the AF is effectively focusing at f/5.6. If one takes the shot at f/1.2 focus moves forward (front focus), and the result is quite blurred. One could use the AF fine adjustment to correct for this, but the correction would be valid only for one aperture. If one adjusts for f/5.6, the image at f/1.2 will be front focused (focused closer than the intended focus point).
He says that the f/1.2 lens is best focused by eye or with live view. If one is focusing by eye with the Canon f/1.4 lens when taking a portrait where the eyes should be in sharp focus and shooting at f/4, the focal point will move backward by an inch or thereabouts, so one could focus on the nose to obtain sharp eyes.
Your comments would be appreciated.
Regards,
Bill
-
Bart,
Your post brings up another problem with AF fine focus adjustment: many large aperture lenses exhibit focus shift when stopped down due to spherical aberration. This can be pronounced with f/1.2 lenses. The peripheral rays (those farthest from the optical axis) intersect closer to the lens along the optical axis than those closer to the optical axis, and the focus shifts further from the lens as the lens is stopped down. I'm not sure what you mean when you say AF is determined wide open.
Hi Bill,
What I mean is that the AF determination is done at the widest aperture of the lens, not at the aperture one has dialed in for the actual subsequent exposure.
The AF optics are a compound lens system, sampling the image with different apertures and hence more or less contribution from the more peripheral rays at the widest aperture, but limited to f2.8 maximum. Here's (http://www.dpreview.com/forums/post/37941968) an old post on DPreview, therefore may be dated a bit, but it explains the principle well.
Here (http://www.dpreview.com/forums/post/50749627) is a more recent post from the same poster, where he clearly states that there are basically 2 AF separator lens apertures in play, f2.8 and f5.6. I know him to be a very well informed person, he used to design imaging sytems for the UK military.
DigLloyd has a good section on focus shift in his Making Sharp Images (a pay site, but well worth the modest cost). He says that Canon AF sees the image effectively at f/5.6 and does not utilize the peripheral rays. He says Nikon is somewhat better in this regard, but does not specify by how much.
Well, that contradicts what the other poster said, f2.8 is most certainly also involved on Canons, and it apparently has been for a long time ... I'm pretty sure that "Its RKM" has a better understanding of the technical principles involved. DigLloyd is known to be a Nikon/Leica/Zeiss lover, so I'm not sure how unbiased his comments are about other brands.
Cheers,
Bart
-
It seems that the days are long gone when photography was a visual art, dependent largely upon soul and intuition.
All this technobabble really makes me sad.
-
Hi Bill,
What I mean is that the AF determination is done at the widest aperture of the lens, not at the aperture one has dialed in for the actual subsequent exposure.
The AF optics are a compound lens system, sampling the image with different apertures and hence more or less contribution from the more peripheral rays at the widest aperture, but limited to f2.8 maximum. Here's (http://www.dpreview.com/forums/post/37941968) an old post on DPreview, therefore may be dated a bit, but it explains the principle well.
Here (http://www.dpreview.com/forums/post/50749627) is a more recent post from the same poster, where he clearly states that there are basically 2 AF separator lens apertures in play, f2.8 and f5.6. I know him to be a very well informed person, he used to design imaging sytems for the UK military.
Well, that contradicts what the other poster said, f2.8 is most certainly also involved on Canons, and it apparently has been for a long time ... I'm pretty sure that "Its RKM" has a better understanding of the technical principles involved. DigLloyd is known to be a Nikon/Leica/Zeiss lover, so I'm not sure how unbiased his comments are about other brands.
Cheers,
Bart
Bart,
Good information, as usual. I looked at the DPReview thread you referenced, and the information does seem valid and the article by Doug Kerr referenced in that thread also mentions that Canon uses two apertures for auto focus. If I understand this correctly, auto focus with his f/1.2 lens would use the f/2.8 set and there would still be some focus shift, but not as severe as would occur at f/5.6. I sent him an e-mail referencing this thread in case he would like to comment.
In any event, these complications should be of interest to readers of this forum.
Best regards,
Bill
-
It seems that the days are long gone when photography was a visual art, dependent largely upon soul and intuition.
All this technobabble really makes me sad.
Photography has always had a technical aspect since photographers had to coat their own plates. There's never between a time when creating high quality photographs didn't need an understanding of the technology involved.
If you'd prefer technology free photography go and post to Instagram.
-
Yes, I could do that. However, I've already had an unsatisfactory experience doing something similar with Canon equipment a few years ago, so I'm a bit reluctant to risk wasting more of my time.
I was very impressed with the sharpness of my Canon EF-S 17-55/F2.8 zoom, until I discovered that it wouldn't accurately autofocus at close distances when set at F2.8 on my new Canon 40D. I sent both lens and body to Canon service under warranty, to get the problem fixed. They were unable to fix it and claimed that both body and lens were within specification.
The Canon 40D didn't have the AF fine-tune feature, so I was stuck. But within a short period, Canon released an upgrade which did boast the AF fine-tune feature, the 15mp 50D. So I bought it, not just for the AF fine-tune, but that was a major attraction.
The irony is, the 50D focused perfectly at maximum aperture with the EF-S 17-55/2.8, without any AF fine-tune adjustment at all. I can only assume that the AF fine-tune has a dual purpose. To compensate for any slight maladjustment of the lens, and/or to compensate for any slight maladjustment of the camera body.
Both lens and body could be within specifications, that is, within the acceptable margin of error, but there could be situations where the errors in both body and lens are at the same end of the allowable spectrum of error, and have an additive effect.
Conversely, there could be other situations where the errors in both the lens and the body are at opposite ends of the spectrum and cancel each other out, creating the false impression that one is lucky to have an ideal camera body and lens without error.
That's my understanding too.
Only regarding error on the mount. Remember that all those pieces of glass have to be set to very fine tolerances all of which have an error. More elements, more errors. If it is a high end lens with fine quality control more elements can be more corrections as mathematically designed. In the real world...
For example I bought an old lens cheap on ebay many years back that is officially a dud. A Minolta 135 2.8 It wont focus at infinity. It is absolutely marvelous for up close work with a depth like a Leica. My guess is the factory made a batch with testing on portrait distance subjects. There was a big oops. I can 'fix' it by moving the group held by a heavy tape inside. I fear I will lose the uniqueness of the output so I keep it for special uses. All my other lenses can take the job at infinity.
-
Only regarding error on the mount.
Why do you presume that, Fine_Art? This is out of my area of expertise, but my general understanding is that the Phase Detection autofocus system does not measure the light at the plane of the sensor, but diverts a portion of the light which has reached the mirror, to separate AF sensors located elsewhere.
I would imagine that the precise position, orientation and distance of those AF sensors within the camera body could affect the accuracy of autofocusing. But I'm just making a reasonable assumption based on very limited knowledge of the process.
-
I would imagine that the precise position, orientation and distance of those AF sensors within the camera body could affect the accuracy of autofocusing.
Hi Ray,
That's one of the variables on the camera body side of the equation. The larger variable may well be the (secondary) mirror that deflects the AF portion of the incoming light. That mirror's position is determined by the main mirror, on which it hinges. Displacement of that mirror will lengthen/shorten the optical path to the AF sensor, and thus the apparent focus position for optimal phase contrast.
And then there is also a potential prism or mirror at the bottom of the mirror box which deflects the light further down the mirror box. There might even be some dust there, throwing the signal levels a bit off maximum achievable.
Lots of small variations/tolerances may add up and/or cancel each other out. And on the image side of the lens, small differences make big deviations.
Cheers,
Bart
-
Hi Bart,
Thanks for confirming that. The fact that my Sigma 35mm doesn't require any AF fine-tune adjustment on the D800E, whether used close-up or at infinity, but does require a +20 adjustment on the D7100, implies there's something very different in the internal adjustments of the D7100 body, compared with the D800E
If the D7100 were to require such extensive adjustments with my other Nikkor lenses, then that would imply that the D7100 body were at fault, or outside the tolerance range. But the fact is, the other lenses I use, such as the Nikkor AF-S 80-400/F5.6 and the 24-120/F4 require very little adjustment on the D7100.
For example, I've got the AF fine-tune on the D7100 set to +5 for the the 80-400. On the D800E the 80-400 doesn't appear to require any AF fine-tune at all, at least at 400mm.
-
this takes very little time ... maybe 20 minutes? even a complex system like a tech camera which requires manually calibrating each lens to the body/back combination takes about 15 minutes per lens. This is once for each camera lens combination... certainly a pretty insignificant amount of time to invest in quality for something you will most likely use for quite some time (often years). I don't think it will infringe on much shooting or family time.
And if you don't want to take the time, many good camera stores offer it as a service. At my store, we will do it for you for $45 for the first lens, $25 for each lens done at the same time. Just drop it off and pick it all up the next day.
Which technique do you advise to do the calibration?
Do you test at two distances?
-
Why do you presume that, Fine_Art? This is out of my area of expertise, but my general understanding is that the Phase Detection autofocus system does not measure the light at the plane of the sensor, but diverts a portion of the light which has reached the mirror, to separate AF sensors located elsewhere.
I would imagine that the precise position, orientation and distance of those AF sensors within the camera body could affect the accuracy of autofocusing. But I'm just making a reasonable assumption based on very limited knowledge of the process.
Your post was talking about lenses so I was talking about the errors in the lens. Yes, you are right the errors in the body will impact the AF as well.
I don't think displacement of the mirror, assuming it stays at 90 degrees, would have an impact. Only a offset of the portion of the image that hits the sensor. Since you see the same thing through the viewfinder you would adjust to what you want in focus. If it goes out of 90 then there would be an error + on one side of the image and - on the other. If the Af sensors are at a different distance from the sensor plane that would make an AF error.
The body seems fairly simple compared to the complexities in the lenses. The lenses change FL so errors are magnified. The mirror just changes the direction.
-
Which technique do you advise to do the calibration?
Do you test at two distances?
For dSLR I prefer FoCal software. Works very well When focus is critical I use LiveView. Also using LiveView to calibrate AF I think would work very well.
For MFDB, I use infinity because a majority of my images will include infinity as part of the scene. For the tech camera, I calibrate infinity using tethered mode to a computer, and I also find where I lose infinity when I stop down and focus closer ... basically I determine visually with each lens what my max hyperfocal setting that keeps infinity sharp is from f/8 up through f/22. With a tech camera I will use LiveView focus occasionally when something close needs to be critically sharp, but generally I just set the lens on one of these predetermined hyperfocal points.