Canon 5DII and 500L/4.5 still has focus problem

[Frank Kolwicz]

Despite doing a microfocus adjustment at least 4 times and getting the same result each time, I'm still having trouble with focussing behind the subject.

I was out this morning photographing a large Canada Goose on a nest in the snow and low light conditions. I focussed with live view on the screen and could clearly see focus with the 10x magnification, yet the image files clearly show that focus was behind the bird - see image attached.

Frank

[Bob Atkins]

If it's in focus with live view, it must be in focus in the image. Live view is a live view (!) of the image as recorded by the sensor. If the lens is set for manual focus (so it doesn't try to refocus when you make the exposure), there's no way focus can change.

So it must be that either (a) The manual focus point wasn't set quite right in the first place or (b) The image is deceptive and focus behind the subject just looks sharper because there's more detail there than in the main subject.

In theory, if the lens is being used stopped down, it's possible (theoretically) that the optimum focus point might change slightly with aperture, but that's only true if a fairly large amount of spherical aberration is present - which is certainly not the case with the 500/4.5L. The chance of this being the answer is very, very remote, even if you were shooting at f16. I presume you were at f4.5 (or close). Focus is always done at f4.5 (or whatever the maximum aperture of the lens is).

[Frank Kolwicz]

I was using MF and at f/8 with the 1.4x converter. I tried to attach a jpg to the message, but I guess it didn't go, it would have shown that there is good focus in some grass just behind the bird while the side and back of the bird (that I foucssed on) was soft. There were water drops on the bird's back that made focus very easy to see and I did this individually for at least a dozen of the 42 images in that series and still got only 3 in focus. My eye is fine and it was clearly obvious that I had achieved focus.

There was no wind, but it was dark, requiring ISO 1600 at 1/40 sec. I carefully watched the LV at 10x and could see no image movement and the sharp grass proves that it worked. The bird was often very still for a second or two or more and the good shots were fine, so it's not blur, it's wrong focus.

This has been an on-going problem that I thought I had solved a couple of times and noted in previous messages to this forum.

I'm going to run a comparison of the 5dii and Xti tomorrow and see how they compare.

I tried to attach a sample image here.

[Bob Atkins]

Images appear here as links rather than in line, but I see what you mean.

I guess I'd try shooting with the lens wide open, just in case there is some sort of weird focus shift going on. You'd probably be better shooting some fixed photographic targets. Maybe a series of small objects at slightly different distances, making sure you are focusing on the center one. 

Focusing in Live View should work without problems. I use it all the time with my 7D when I'm doing lens testing and it gives perfect results as far as I can tell. I've used it with dozens of lenses.

[bmpress]

Next time use mirror lock. Vibration might be the culprit here.

[Frank Kolwicz]

I guess you missed the reference: Live View is MLU and the problem is that the image shows good focus *behind* the goose, not image blur, but thanks for your reply. I also used a remote release, viewed the subject at 10x in Live View and waited for all sign of movement to stop.

Also, in response to a previous comment: the goose was not moving, it was on a nest and had only stood up to make some "housekeeping" adjustments.

[Frank Kolwicz]

I know this is supposed to be impossible, but I've now convinced myself, against all opposition, that it's happening:

I've just done another (of nearly a dozen!) series of tests using the 5dII, 500L/4.5; tripod, this time with a Manfrotto "fourth arm" to stabilize the camera; manual focus at 10x with Live View and remote release and the actual focus from the memory card image is *behind* the subject I focussed on, by a considerable amount. I have photos of the test setup (a backyard situation typical of how I use the camera, not a test bed) and the resultant images, about a dozen of them. AF microadjust is turned off for some of the images. If you want to see them, I'll post them and send you the link.

The images aren't all exactly the same, there's a tiny bit of drift, which may be due to my manually refocussing between shots, but it is consistently behind the subject (a rock in a pile of rocks) and remarkably in critical focus.

The only possibility that I can think of is that the focussing motor moves when the shutter is tripped (even MF is done with the motor with this lens, there is no mechanical connection). I've done this with the 1.4x telextender and a pin-hacked 1" extension tube and there's no difference.

I tried insulating all the camera/lens readout pins, but that disables focus entirely, it needs some power to the lens for focussing. If anyone knows how to keep power to the lens, but block AF, please let me know, so I can try that and see if there's some kind of interference from the AF circuit.

[Bob Atkins]

I can't explain your results, but here's the pin info. I don't know if the lens will focus with the lens data pins blocked off, but the data appears to be on pins 5 and 6.

It's possible that focus commands go from the lens to the body and then back to the lens. If so if you block off the data pins you will lose focus even though the lens itself will have power.

What you might be able to do is to get focus, then partly unmount the lens, i.e. push the release button and rotate it by maybe 10 or 15 degrees.  That should disable all connection, but keep the lens properly registered with the flange, i/e. the focus should not change. I suppose there is some theoretical risk of damaging something, but I never power the camera off when changing lenses and I've never had a problem caused by rotating the lens with the power on the camera.

Pinout and pin functions:
Left to right, looking at the front of the body:
1. VBAT
2. P-GND 3. P-GND (pins 2 & 3 are common on the lens)
4. VDD
5. DCL
6. DLC
7. LCLK
8. D_GND

Pin Functions
VBAT - Motor power
P_GND - Motor power ground
VDD - Logic circuitry power
D_GND - Logic circuitry ground
DCL - Data from body to lens
DLC - Data from lens to body
LCLK - Clock

[Frank Kolwicz]

Thanks, Bob, I've tried doing the trick with disengaging the camera by giving it a slight rotation, but it then does not stop-down to my working aperture, f/8 and turns Live View off, so no mirror lock up.

Making the comparison of engaged/disengaged at f/4.5, there is no visible shifting of the focus , but focus is soft enough that it is hard to say.

Comparing f/4.5 with the camera rotated to f/8 in the normal position, the image at f/8 is much sharper and obviously about 1/2" or so behind the point focussed on. Of course DOF is much deeper so I'm estimating the distance to middle of the sharp zone which extends from the focal point back.

I'm going to try focussing stopped-down next (if the sky brightens up enough for me to see it well) and work on the pins later.

[Frank Kolwicz]

Well, Canon wins this round, too. It turns out that you can't use the Preview button on the side of the lens mount and focus at the same time while in Live View! Who'd a thunk it.

You CAN focus and preview while using the optical viewfinder (no Live View) and I did a series that way, judging focus by way of the AF confirmation light and the center of the DOF is always behind the focal point, as estimated from the texture of the smoothly curved rock surface. There is a slight variation due to the AF system guessing which part of the rock has the highest contrast within the AF sensor (middle one only). The curved rock is a good simulation of my normal subjects.

My similar vintage EF 100-300L in close-up mode (at f/8 and 300mm and at the minimum focus distance to exaggerate the DOF to a similar level as the 500L) does not move the focus after manually focussing in Live View. In MF this lens does mechanically couple the focus ring with the moving element, unlike the 500/4.5L.

Has anyone repeated my tests with their supertelephotos to see if this is just a defect in my camera or lens?

[Frank Kolwicz]

I've been running these tests since about 11AM, it's now 3 and I'm tired.

The last thing I just finished was with pins 5&6 insulated with teflon tape. This has cut off the normal AF mode, but not affected AF in Live View (I forget which is which technology) and it looks like focus has stopped shifting, based on only 3 trial frames. I'm going to take a rest and then go out and do some birds with the lens set up with the taped pins and see what happens.

[Frank Kolwicz]

The birds were not cooperative yesterday, but I did a couple of static subject tests with a high contrast object on a tilted flat surface (purple plastic bottle cap on a plastic box with raised pattern and letters) as well as some natural objects in the yard (flowers and leaves).

It's not as obvious as before, but the center of the "sharp" zone is still behind the point I focus on - no matter how carefully I focus. This is easiest to see in the plastic cap image where the lettering on the box clearly shows that the front of the sharp zone is in line with my focal point and recedes from there.

[Bob Atkins]

If it's actually refocusing you might be able to see it. Look at the distance scale with something like a 10x loupe and see if you can see it "jog" a fraction of a mm when you release the shutter.

If it's not refocusing then you must be seeing some sort of focus shift when the lens stops down. This would be very surprising for a 500/4.5L. It normally only happens with very fast lenses. It wouldn't be surprising to see it with a 50/1.0, 50/1.2 or an 85/1.2, but it would be pretty unusual to see it on a  lens that was f4 or slower. It's due to spherical aberration and that really isn't an issue with lenses as slow as f4 or even f2.8. It's the superfast f1.4, f1.2 and f1.0 lenses that you might look for it with. Normally it's only an issue when stopping down from wide open to f2.8 or f4. By the time you get to 5.6 or f8 the increased DOF usually means the focused point is sharp.

If you think that the lens isn't shifting focus, but that stopping down is affecting focus you could compensate with microfocus adjustment. If you know you're always going to be focusing at f4.5 and shooting at f8 you could tweak the microfocus adjustment until you get the f8 images in focus. Of course then your f4.5 images will miss exact focus...

[Frank Kolwicz]

I did try some additional tests.

With an Xti body on the lens it worked fine, so I guess that means it's not aberration, but something to do with the 5dii.

I like your idea of tweaking it for f/4.5 focus/f/8 exposure and I'll try to set it up that way as f/4.5 is not acceptably sharp, so I never use it that way, but deal with the bit of noise at higher ISO instead. 99% of my shots are at f/8 nominal, ISO 1600 and with a 1.4x converter.

I had originally set microfocus according to the book (f/4.5 focus and taking aperture) and that required an offset of -12 to -15. I've redone it with the taking aperture at f/8 and it requires no correction, so there's a big difference for some reason and I bet that's what was driving me nuts. Now I'll have to check some other apertures to see if there are more differences.

Only time will tell now that we're getting into migration again and I'll have plenty of subjects to work with.

Thanks with you help and your patience.

[Bob Atkins]

It may be significant that the XTi has no microfocus adjustment as far as I know.

I suppose it's possible that the microfocus adjustment is kicking in when it's not wanted? In live view though it shouldn't be using any microfocus adjustment since live view uses contrast detection from the sensor image. I think microfocus adjustment is a correction for the phase detection AF scheme used in normal DSLR optical viewing mode. There would seem to be no reason why it should affect contrast detection AF, and in fact it probably shouldn't.

I'm surprised you don't find f4.5 acceptably sharp at f4.5.  The 500/4.5L is normally pretty sharp wide open.