p.2 #1 · 5D3 outside AF point issue with wide lenses
gpchase wrote:
I just shot my brick fireplace from 10' with all four corner focus points of the lens and the center point. Each point focused on bricks and every file appeared okay although the center seemed sharper overall..??
Center of the lens is the sharpest part... remember, the 5DM3 focus points are the furthest from the center of any canon camera. Thus, the IQ is going to be impacted by the general lens falloff of sharpness.
p.2 #2 · 5D3 outside AF point issue with wide lenses
Yes, center lens performance will be best. So expect it to be softer closer to the edges. The key here is to watch if the focus scale changes. The 24-70 may not show much IQ difference like the 24L, 35L and 50L.
p.2 #6 · 5D3 outside AF point issue with wide lenses
Nothing new here... all my Canon bodies (5D, 5D2 and 1Ds3) have shown unreliable behavior on the outer AF points with (fast) wide angle lenses / focal lengths. For example: 35L and 24-70L. The 24-105L performs better, but this could be because DOF @ f/4 is covering things up. My educated guess it is due to the negative effects of (heavy) field curvature on the AF. Nothing 'wrong' with your camera... it is just the way it is. Stopping down helps to improve outer AF point performance with wide angle lenses.
p.2 #7 · 5D3 outside AF point issue with wide lenses
yurch wrote:
My 35L is fine, but I have the same problem with Sigma 50 1.4. Naother user has this problem with 50L.
My 35L appears to be terrific with the 5D3. 24L II needs some MA, but also seems to behave. I'm the one with the 50L that is all over the map. It's going to Canon for a checkup.
p.2 #8 · 5D3 outside AF point issue with wide lenses
same issue with mine.
Discovered this on an actual shoot - environmental portraiture , when I was placing the model in lower right area of the frame. Just could not get it to be sharp even when i fired multiple times. I did tests on the tripod at home and my 50L needed a +7 and my 35L needed a +5.
This makes the center a bit softer , but i rarely take photos with the subject at the exact center. its something that I can live with.
The 50L is still iffy, but its the 50L , what do you expect - I never go to that lens when I need sharpness anyway.
85L , zero issues whatever AF point i use, near or far, always locks on - sharp at 1.2. focusing is perfect. I never thought Id say this about the 85L .85L + 5D3 is an dream combo.
p.2 #9 · 5D3 outside AF point issue with wide lenses
Invertalon wrote:
How I would recommend testing:
1.) Place camera on tripod. Aim at high contrast target with little room for focus errors.
2.) AF using center point. Watch focus scale on the lens. De-focus and let aquire focus a few times, watch to see where the lens ends up on the focus scale after a few times. For example, let's say it lands at the 3ft mark on the scale.
3.) Shift tripod head without shifting distance from target. Align target with outer AF point. Once situated, defocus and let the camera acquire focus. Do this a few times and check out the results. Remember, greater distances seem to show the problem more then closer subjects, although I saw it with both.
4.) In my experience, the center point showed that 3ft mark with center, and would shift left over to about 2.5ft (just an example) with outer AF point. Showing the body is front focusing. Now remember, if testing with a 24-105 or 24-70, etc... Due to the smaller apertures, you won't see the results as clearly as you will faster prime lenses. Which is why the 35L shows it so well. ...Show more →
I don't want to be a pain about this but this procedure, as written, is not enough to automatically assume the camera is not focusing correctly. You have to observe the sharpness at the focus point, either with live view or with the image outside the camera. Some of the lenses, particularly the wide ones, have significant field curvature. I could not find any specific articles on the 35L but the 35 f2 has tons of field curvature and I suspect the 35L has enough to be noticeable in controlled situations.
In the second shed picture you posted, it does appear that most of the image is soft but I still would like to know where in the picture the outer-focus point was positioned. If it was located on a lower-contrast portion of the image, that might also have an impact.
I just think it is important to consider field curvature in your testing and make sure you are accounting for it before jumping to the conclusion that the 5D3 autofocus is bad on the outer points. Maybe you have done this but based on how the information is being presented in this thread (with the emphasis on how the focus ring moved), I have doubts.
p.2 #10 · 5D3 outside AF point issue with wide lenses
I was using focus point expanded, so if there was not enough detail it would go where there was. I repeated the same test on many other subjects with identical result. It was not a one off test. It does appear to be the effects of field curvature with the AF system and wide lenses, but you think Canon could counteract it somehow.
Secondly, using live view at the same spot as the outer AF point, the resulting image is very sharp. Only when the camera attempts to AF is it soft and shifts the focus scale.
I know for a fact that it is not due to poor testing. It is a real issue if you enjoy using outer points with fast, wide lenses. Now I am not saying the AF is faulty on the 5D3, it is amazing. IF you don't use outer points with wide lenses.
p.2 #11 · 5D3 outside AF point issue with wide lenses
I tested outdoors at 10 ft distance. No problem with 24 L, 35 L, 50 L. No movement of the focus ring when changing between center point and outer point. No discrepancy with live view focusing.
I did detect the "issue" with 35 L and 24-105 in a quick indoor test at ~4 feet last night, but I'll redo that more carefully before drawing conclusions one way or another.
p.2 #13 · 5D3 outside AF point issue with wide lenses
j5daniel wrote:
The 50L is still iffy, but its the 50L , what do you expect - I never go to that lens when I need sharpness anyway.
PhilDWedding wrote:
I must have the only good copy of that lens... it's never let me down. Not once. And it was secondhand.
Nice try, Phil, mine is a stellar copy, too. I just have to be careful in selecting the AF point and minimize recomposing. My copy is also used, I meant, previously owned...
p.2 #15 · 5D3 outside AF point issue with wide lenses
Invertalon wrote:
The key here is to watch if the focus scale changes.
Invertalon wrote:
Since the shift will be relative to whatever the camera is doing anyway. Just watch the focus scale with both, not relying on IQ
Just because there are focus scale changes, it does NOT mean there's an Auto Focus problem with field curvature. Remember, the focus scale is calibrated for the center of the lens. The scale assumes you are using the center AF point. The lens itself has no idea which AF point you're actually using.
If you're using an outer AF point at close focusing distance -- and there's field curvature -- then that distance scale will NOT be telling you where the actual focal plane is. It will tell you where the focal plane of the center point would be. On a lens with field curvature, that would be behind the focal plane of an outer AF point.
Therefore, if you focus on a flat plane with center point, the scale will show you a certain distance. If you use an outer point to focus on the same plane, the distance scale is SUPPOSED to move further back if there's field curvature.
p.2 #16 · 5D3 outside AF point issue with wide lenses
I wonder what the result would be if you do MA with an out focus point and focus with the same point. This test should tell you if the mis-focus is caused by the lens curvature.
p.2 #17 · 5D3 outside AF point issue with wide lenses
^
MA'd with my outside point, it requires about +9 or +10... Center point requires +4. So it is different by using different points. Which makes sense... Looking at photozone.de, the 35L has about -1.5% barrel distortion. So if you shift the center AF point over to the edge, your plane of focus stays put (while the subject is distorted behind), causing the front focus. That is why higher MA values (shift focal point back) correct the distortion you get towards the edges. Starting to make more sense. The 50L shows about the same level of distortion at about -1.45%, so it should show similar results. The 24-105L has much more at above 4%, but since the dof is much larger you don't notice as much. Here is the chart for the 35L:
But my question is, why would the camera's AF system regardless of AF point see beyond the lens curvature? Would it not lock focus regardless of the distortion of the lens? So even if the barrel distortion causes the image to actually fall behind the center point focal plane, why would the edge AF point not focus back to ensure sharp focus? Instead, it seems to move forward?
I will just focus/recompose with subjects farther away since at larger distances it front focuses quite a bit. But if you stop down, the issue goes away... So only a real hassle (if a hassle at all, focus and recompose with the 35L is very easy due to larger DOF).
I guess if it was an outer point issue it would happen at any FL with any lens. But it consistently only does it with wider FL. Still sucks, but I guess you can't really do anything about field curvature.
p.2 #18 · 5D3 outside AF point issue with wide lenses
It might be hard to predict how field curvature affects focus at different distances. Canon might be able to develop a firmware to compensate the difference if they want.
If you mainly use outside points, it might be good to MA with one of the points you use most often. I would assume that a L lens should have the same degree of field curvature at all the symmetric points.
p.2 #19 · 5D3 outside AF point issue with wide lenses
definitely its not an outer point AF issue.
longer lenses starting with 85L , AF perfectly even with outer points.
Maybe in the future, canon should include automatic field curvature AF correction with their lenses - for Full frame bodies.
I believe its not an issue with crop sensors (when using FF lenses).
It wasn't noticeable with 5D1 and 5D2, coz we never cared about outer AF points anyway.
They couldn't lock in to begin with and when they did lock ,the focus was off . We just assumed it was because the AF points were crap but apparentl it may have been caused by field curvature instead.
Invertalon wrote:
^
MA'd with my outside point, it requires about +9 or +10... Center point requires +4. So it is different by using different points. Which makes sense... Looking at photozone.de, the 35L has about -1.5% barrel distortion. So if you shift the center AF point over to the edge, your plane of focus stays put (while the subject is distorted behind), causing the front focus. That is why higher MA values (shift focal point back) correct the distortion you get towards the edges. Starting to make more sense. The 50L shows about the same level of distortion at about -1.45%, so it should show similar results. The 24-105L has much more at above 4%, but since the dof is much larger you don't notice as much. Here is the chart for the 35L:
But my question is, why would the camera's AF system regardless of AF point see beyond the lens curvature? Would it not lock focus regardless of the distortion of the lens? So even if the barrel distortion causes the image to actually fall behind the center point focal plane, why would the edge AF point not focus back to ensure sharp focus? Instead, it seems to move forward?
I will just focus/recompose with subjects farther away since at larger distances it front focuses quite a bit. But if you stop down, the issue goes away... So only a real hassle (if a hassle at all, focus and recompose with the 35L is very easy due to larger DOF).
I guess if it was an outer point issue it would happen at any FL with any lens. But it consistently only does it with wider FL. Still sucks, but I guess you can't really do anything about field curvature.
p.2 #20 · 5D3 outside AF point issue with wide lenses
Incidentally, the 5D2 I own has a similar issue..with all lenses. Outer points miss pretty consistently in the same fashion...
Irony is, when the 5D classic outer points don't hunt, they are more accurate than the 5D2's outer points.
Invertalon wrote:
Just wanted to share my findings today to see if others experience the same issue.
At first, I thought it was limited to my 35L, but does the same exact thing with my 24-105L. The shallow DOF of the 35L just shows it much more.
I calibrated my 35L to my body using Focal... Extremely reliable AF performance using center point. Never misses. Switch to any of the outer cross AF points and lens front focuses quite a bit. You can see the DOF scale shift over... Checked on multiple targets at various distances on a tripod (so distance does not change) and consistently the DOF scale shifts to the left once I switch to outer AF points. Resulting in soft shots wide open using outer AF points at large apertures.
I thought it was just limited to that lens, but then I tried it with my 24-105... Under 50mm, it shows the exact same behavior. Over 50mm, it does not appear to do the shift.
Almost like with wider lenses, the outer AF points are having trouble locking on. I plan to try my 135L next to see if that does it as well... If my theory is correct, it will not shift due to the FL. And from my shooting with it so far, have not seen this happen either.
Just wanted to share this with everyone though. I plan to send to Canon if firmware does not address this issue with the AF.
