[jeffkrol]

Ran into this thread regarding, in part , angled focus targets. Seems these are a no-no, at least for Canon....
Do not attempt to autofocus on an angled chart, because doing so will degrade the consistency of the camera's focusing measurement. Keep in mind that the camera's AF sensor is comprised of multiple pairs of linear pixel arrays. If you attempt to autofocus on a single line in an angled focusing chart, only a few pixels from each active pixel array will "see" the target. Ideally, the contrast in the reference target should cover the entire area of the camera's center focusing point, and the reference target should be perfectly parallel to the camera's focal plane.
News: LensAlign™ Seen! Announcement coming soon! - Page 3 - Open Photography Forums
Other interesting stuff here as well.....
Make note that I'm giving RH creditfor, I believe, mentioning it earlier (not to use angle tests). Call it a Christmas miracle.........

[dazman]

Useful info...for me. I get inconsistent results from focus charts. Another member (Heliphoto) suggested a different method which I will be trying this weekend using dvd covers...sorry can't find the link.
Thanks for the info, Jeff

[GeneV]

Hmm. It seems to me that objects in the real world have angles. That is the challenge for AF, IMHO.

[jeffkrol]

Hmm. It seems to me that objects in the real world have angles. That is the challenge for AF, IMHO.

But few that I can think of at 45 degrees or greater. Even shooting prone people, your usually not at 45 or greater. You'd likely hit the nose and be on plane again
Anyways some more info on "Lensalign"
Digital Photography - RawWorkflow.com - LensAlign?
Luminous landscape

[GeneV]

Maybe I'm not getting it, but it seems we take photos looking up and looking down. Objects in rooms are at all kinds of angles. Planes, trains and automobiles have angles. Even noses are at something like 45 degrees. What am I missing?

[Class A]

The service manual for the Samsung GX-10 shows a setup in which the focus target (full black to full white transition) is parallel to the sensor while the measuring area / scale is angled at 30 degrees.

Beats me how the sensor can see whether a fully black or white area is angled but apparently it matters sufficiently to warrant the slightly more complicated setup. I guess it has to do with lighting issues.

BTW, chart camera distance is 1m and the lens to be used is an FA 50mm.

[Lowell Goudge]

aside from the technical issue noted here about focusing on an object that is at right angles to the axis of the lens, what bothers me about all the focusing chart tests I have seen posted here is that I have NEVER seen a photo of one where the camera (from left to right) is square on the target. this makes the distance scales on both sides not on the focusing plane, and really has only one point (not a line or a surface) at the correct focusing distance, and then people complain about front and back focusing.


The measurements being made, unless done carefully, can lead to all sorts of problems.

until I see some really properly done shots, I will generally ignore most focusing tests

[Kerrowdown]

I use the old fashioned focus test, what does the picture look like.

[GeneV]

I use the old fashioned focus test, what does the picture look like.

I posted the same thing a few days ago regarding the brick wall test.

Last night I was looking at some daytime street photography I shot with the DA 50-200mm with lots of detail in the background. That lens is decent, but not reputed to set the world on fire with its resolution. However, it was hard to find a shot in which the lens got in the way of a good photo.

[Marc Sabatella]

The reason to shoot a test at an angle is a simple practical one - it's the easiest way to set up a shot in which if the target is in focus, there will be thing in front of and behind the target that can be used to check whether the focus was missed and if so, by how much in which direction. If you held th chart up vertical and shot it straight on, there would be nothing in front of or behind the chart to judge where the focus actually was. All you could do is say, "yeah, that probably looks sharp enough", or "hmmm, I think it should be sharper".

I don't understand enough about how AF works to say if shooting at an angle really causes problems. But if you, you'd need to find some other way around the issue mentioned above - you need a way to evaulate the reuslts, and shooting a flat target at an angle is the most obvious way to get that.

And Lowell - I compeltely agree that most people trying to perform focus tests are doing so very ineffectively. But of course, it's that much worse when they *aren't* using any sort of controlled target like a chart but merely post a picture of a mass of people and say, "see, I was focusing on my daughter, but her friend in front of her is more in focus - clearly, my camera has an FF problem!" The charts at least remove the variable of *what* the camera was actually trying to focus on.

But they are still trickier than hell to get reliable results from. Especially when one also consider that most lenses don't have a perfectly flat "plane" of focus. So even if the lens focuses on the center target perfectly, and you set up the chart perfectly square, it would be *absolutely correct* for the scales on the side of the chart to show the focus being a bit forward of the center line. How much forward? I have no idea - it's lens dependent, of course. I suppose running the test using MF would give you an idea of what to expect, assuming you were confident about the accuracy of manually focusing on a center target - which at large apertures is *not* a given *at all* since the viewfinder DOF is so much greater than the actual DOF. Better perhaps to use a different version of the test with more stuff in the middle as well as on the sides, so you can judge where the focus really is at the center as well as along the sides.

[Michael Barker]

The reason to shoot a test at an angle is a simple practical one - it's the easiest way to set up a shot in which if the target is in focus, there will be thing in front of and behind the target that can be used to check whether the focus was missed and if so, by how much in which direction. If you held th chart up vertical and shot it straight on, there would be nothing in front of or behind the chart to judge where the focus actually was. All you could do is say, "yeah, that probably looks sharp enough", or "hmmm, I think it should be sharper".

Not much you can do with a single shot, but why not take your first shot, then bracket focus in front and behind of where the camera chose, review and decide which is best?

[GeneV]

The reason to shoot a test at an angle is a simple practical one - it's the easiest way to set up a shot in which if the target is in focus, there will be thing in front of and behind the target that can be used to check whether the focus was missed and if so, by how much in which direction. If you held th chart up vertical and shot it straight on, there would be nothing in front of or behind the chart to judge where the focus actually was. All you could do is say, "yeah, that probably looks sharp enough", or "hmmm, I think it should be sharper".

I would totally agree that this is the only way to judge front or back focus.

My concern is really with the non-angled test shots which turn out to be angled. I have gone so far as to put a long aluminum carpenter's level across the front of the lens, and measure from each side--a two person operation--to get the ccd parallel to the wall I am shooting to judge edge sharpness. (This only works with flat walls covered with test targets) There are just so many variables when differences in the degrees of sharpness are small.

[jeffkrol]

Actually this seems like a superior means to test f-B focus..... A bit pricey though.
NO I get no commission or any gratuety for pimping this thing.

I'll wait til the "NAME YOUR FAVORITE COUNTRY" does a cheaper Knockoff

[Marc Sabatella]

Not much you can do with a single shot, but why not take your first shot, then bracket focus in front and behind of where the camera chose, review and decide which is best?

You certainly could, but that makes the results a lot harder to quantify, and still leaves a considerable amount of guesswork (just how much in front to try bracketing to avoid accidentally over-correcting etc). It's just a heck of a lot more imprecise than a test that can show you a gauge you can read clearly to see exactly how much FF or BF there is.

And yes, Jeff- sorry, I didn't check out the link. Looks like a nice device, but I guess I remain unconvinced that the angle method is that bad, since as others have mention, even if it is not an *ideal* test subject, it does mimic reality. Anyhow, I now quite convinced that the majority of reported focus problems are actually cases or poorly run tests, and given that my own cameras all seem to work just fine, I've kind of lost interest in the focus test business, except to caution others who are starting to get wrapped up into it and are leaping to premature conclusions based on on poorly performed tests.

[jeffkrol]

You certainly could, but that makes the results a lot harder to quantify, and still leaves a considerable amount of guesswork (just how much in front to try bracketing to avoid accidentally over-correcting etc). It's just a heck of a lot more imprecise than a test that can show you a gauge you can read clearly to see exactly how much FF or BF there is.

And yes, Jeff- sorry, I didn't check out the link. Looks like a nice device, but I guess I remain unconvinced that the angle method is that bad, since as others have mention, even if it is not an *ideal* test subject, it does mimic reality. Anyhow, I now quite convinced that the majority of reported focus problems are actually cases or poorly run tests,

I fail to see how a thin black line on a piece of paper "mimics reality"
Anyways if Canon says "no" and Doug Kerr is seemingly in agreement, that's good enough for me............
THESE ARE QUOTES!
Originally Posted by Doug_Kerr View Post
Regarding the matter of using just an angled test target for AF testing, I happened to run into this from Chuck Westfall of Canon in his Tech Tips column in The Digital Journalist for December 2008:

• Do not attempt to autofocus on an angled chart, because doing so will degrade the consistency of the camera's focusing measurement. Keep in mind that the camera's AF sensor is comprised of multiple pairs of linear pixel arrays. If you attempt to autofocus on a single line in an angled focusing chart, only a few pixels from each active pixel array will "see" the target. Ideally, the contrast in the reference target should cover the entire area of the camera's center focusing point, and the reference target should be perfectly parallel to the camera's focal plane.

This is consistent with Canon recommendations I had seen in the past.

Here is the link to Chuck's column:

http://www.digitaljournalist.org/iss...tech-tips.html



Best regards,

Doug
AND...........
Here are a few additional precautions to observe:

• Do not attempt to autofocus on an angled chart, because doing so will degrade the consistency of the camera's focusing measurement. Keep in mind that the camera's AF sensor is comprised of multiple pairs of linear pixel arrays. If you attempt to autofocus on a single line in an angled focusing chart, only a few pixels from each active pixel array will "see" the target. Ideally, the contrast in the reference target should cover the entire area of the camera's center focusing point, and the reference target should be perfectly parallel to the camera's focal plane.

• For best results, manually set the focus on the lens to infinity for every exposure before allowing the camera to autofocus the reference target.

• Expect some minor variations in focusing accuracy within each set of three test images, even though they were all taken at the same microadjustment setting. This is completely normal, and is due to the tolerances of the camera's AF system.

• Expect smaller microadjustment settings to have a greater effect with telephoto lenses, and vice versa for wide-angle lenses.

• If you are attempting to set microadjustments for a zoom lens, it is important to realize that the camera's setting may only be accurate for the focal length setting you test. The instruction book suggests testing at the longest focal length of the lens, but you may find it more efficient to choose the focal length you use most often.

• Some EOS cameras and some EF zoom lenses may require more sophisticated calibration than the in-camera AF microadjustment settings can provide. In such cases, it may be necessary to have calibrations performed at a Canon Factory Service Center.

• Last but not least, there is no "official" Canon method for setting AF microadjustments, so this procedure is unofficial. If you think you can do better, then by all means, go for it. Towards that end, be advised that some independently made tools are designed to help you set AF microadjustments accurately. One of these is the LensAlign kit, due out soon from RawWorkflow.com:


Tech Tips - The Digital Journalist