morganb4 wrote:
Daan, that test you just did, what were the times of day Was one shot under predominantly daylight, whilst the other predmonantly tungsten?
Both were shot in the morning under natural daylight (three big windows in the roof at an angle). But sometimes there were clouds in front of the sun, hence the difference in WB.
Daan B wrote:
BTW Did you do this tests with a 1Ds3? When in focus, you have a sharp 50 at 1.4
No, I did this test a while ago with my 1D (classic).
Just to be clear: with refocusing I meant: take the shot with the same aperture several times, and refocus in between by pressing the shutter halfway. Even better: manual focus the lens to infinity and/or the closest distance between each shot and AF again.
My expierence is that in low light, the AF is acting like this (at least with my 50/1.4). In bright light, I don't have this problem. Maybe this explains the difference between the 2 series of shots with the 35/1.4 you expierenced.
lexvo wrote:
No, I did this test a while ago with my 1D (classic).
Just to be clear: with refocusing I meant: take the shot with the same aperture several times, and refocus in between by pressing the shutter halfway. Even better: manual focus the lens to infinity and/or the closest distance between each shot and AF again.
My expierence is that in low light, the AF is acting like this (at least with my 50/1.4). In bright light, I don't have this problem. Maybe this explains the difference between the 2 series of shots with the 35/1.4 you expierenced.
My aim was to see if there were any differences between AF and MF. I doubt if the differences between the two series of 35L shots was caused by the light. Both were shot in good daylight, not low light. Also, I get focus errors under all lighting. To me it proves the random nature of these focus errors. I think the cause of these focus errors lies within the body itself and not in the shooting circumstances.
lexvo wrote:
Daan, just to be sure: did you take multiple shots when you tested the AF and did you refocus in between? I found that my 50/1.4 is not consistent in low light levels. Several shots with AF give different results. See this example:
These were all at f/1.4, I just pressed the shutter/AF several times. #3 and #4 are OK, #2 is 'almost' OK and #1 is off.
I have been saying this for some time: Canon's AF system is not accurate enough for f1.4 lenses on high megapixel cameras. This behaviour is normal on Canon. With One-shot you can improve performance by pressing focus twice. The first time, the Canon system guesses focus it isn't accurate enough. The second time you press, because focus was nearly right already, you get a more accurate focus. The reasons for this are pretty clear if you read Canon's white paper on how their focussing technology works. Canon needs to rethink it's focussing system for those who wish to use f1.4 lenses on 21 megapixels or more. The good news is that since a double focus procedure works better, Canon may be able to build a better system, possibly even in firmware. But for the moment, I'm using manual focus and focus bracketing at f1.4.
I have the combo and I have found that most of my lenses required AF micro adjustment. Prior to adjusting-- they were all off, which is easily visible wide open and as you see.
I want to add a thought. It seems clear from Canon white papers, etc., that the outer autofocus points are less precise than the center autofocus point. Typically Canon has described the precision of the outer points as within 1 depth of field, but depth of field of course varies with magnification and what is considered acceptable sharpness. It seems to me that with a 21megapixel camera (and the greater magnification this allows) that what we are probably seeing here is that the lack of precision in the outer focus points is likely to create unacceptable results that would not even be within the 1 DOF that Canon claims. So, I suspect we aren't seeing AF hiccups or miscalibration just a general lack of precision in the system. Note that precision is not the same as accuracy. You can be accurate but not very precise. I shoot a gun this way. I will miss three feet to the right one time. Then three feet to the left the next time. Then three feet high the next time. But if you would average how much I miss you would see that the average is still at the center. In contrast my Dad is very precise but not completely accurate. He missed just about every time by about 3 inches to the right. In every day language we mix accuracy and precision, but they are separate concepts. I think Daan has shown that his outer focus points lack the precision that one would hope, but they aren't necessarily inaccurate or needing calibration. If you take the shooting example above. Little can be done for me and my lack of precision, but if you adjusted my Dad's sights just a bit he could be bang on accurate and very precise. In much the same way I doubt calibration is going to help 1DsMKIII owners. You might get a few more in focus, but the basic lack of precision in the outer focus points is the problem--I suspect.
So what is one to do. First, when testing taking multiple shots (like 10 or 20) per setup will help get at the precision issue. You will be able to see if the focus is just all over the place. Second, a work around worth considering if autofocus is necessary is using the higher precision center point and planning to crop if that will provide the print that one wants. Here the extra megapixels of the 1DsMKIII might help. Basically, however, what is needed is for Canon to develop more precision in the AF system and probably to use points like the center point in outer focus point. I hope this helps.
My undrestanding was that the 19 user selectable focus points were all high precission.
Steve Spencer wrote:
Hi,
I want to add a thought. It seems clear from Canon white papers, etc., that the outer autofocus points are less precise than the center autofocus point. Typically Canon has described the precision of the outer points as within 1 depth of field, but depth of field of course varies with magnification and what is considered acceptable sharpness. It seems to me that with a 21megapixel camera (and the greater magnification this allows) that what we are probably seeing here is that the lack of precision in the outer focus points is likely to create unacceptable results that would not even be within the 1 DOF that Canon claims. So, I suspect we aren't seeing AF hiccups or miscalibration just a general lack of precision in the system. Note that precision is not the same as accuracy. You can be accurate but not very precise. I shoot a gun this way. I will miss three feet to the right one time. Then three feet to the left the next time. Then three feet high the next time. But if you would average how much I miss you would see that the average is still at the center. In contrast my Dad is very precise but not completely accurate. He missed just about every time by about 3 inches to the right. In every day language we mix accuracy and precision, but they are separate concepts. I think Daan has shown that his outer focus points lack the precision that one would hope, but they aren't necessarily inaccurate or needing calibration. If you take the shooting example above. Little can be done for me and my lack of precision, but if you adjusted my Dad's sights just a bit he could be bang on accurate and very precise. In much the same way I doubt calibration is going to help 1DsMKIII owners. You might get a few more in focus, but the basic lack of precision in the outer focus points is the problem--I suspect.
So what is one to do. First, when testing taking multiple shots (like 10 or 20) per setup will help get at the precision issue. You will be able to see if the focus is just all over the place. Second, a work around worth considering if autofocus is necessary is using the higher precision center point and planning to crop if that will provide the print that one wants. Here the extra megapixels of the 1DsMKIII might help. Basically, however, what is needed is for Canon to develop more precision in the AF system and probably to use points like the center point in outer focus point. I hope this helps....Show more →
BTW Did you had the opp to do a little 1Ds3 + fast wide angle testing yet?
Not yet mate, still going through the 1000 odd 1Ds Mark 3 shots I took yesterday and it is hard work because as much as I am starting to like the large detailed images, my computer hates them!
So, so far my upgrade cost for the 1Ds Mark 3 is likely to be £4900 for the body itself and £2500 for a top of the range Mac Pro!
As soon as I have done it, I will post, don't worry.
I want to add a thought. It seems clear from Canon white papers, etc., that the outer autofocus points are less precise than the center autofocus point. Typically Canon has described the precision of the outer points as within 1 depth of field, but depth of field of course varies with magnification and what is considered acceptable sharpness. It seems to me that with a 21megapixel camera (and the greater magnification this allows) that what we are probably seeing here is that the lack of precision in the outer focus points is likely to create unacceptable results that would not even be within the 1 DOF that Canon claims. So, I suspect we aren't seeing AF hiccups or miscalibration just a general lack of precision in the system. Note that precision is not the same as accuracy. You can be accurate but not very precise. I shoot a gun this way. I will miss three feet to the right one time. Then three feet to the left the next time. Then three feet high the next time. But if you would average how much I miss you would see that the average is still at the center. In contrast my Dad is very precise but not completely accurate. He missed just about every time by about 3 inches to the right. In every day language we mix accuracy and precision, but they are separate concepts. I think Daan has shown that his outer focus points lack the precision that one would hope, but they aren't necessarily inaccurate or needing calibration. If you take the shooting example above. Little can be done for me and my lack of precision, but if you adjusted my Dad's sights just a bit he could be bang on accurate and very precise. In much the same way I doubt calibration is going to help 1DsMKIII owners. You might get a few more in focus, but the basic lack of precision in the outer focus points is the problem--I suspect.
So what is one to do. First, when testing taking multiple shots (like 10 or 20) per setup will help get at the precision issue. You will be able to see if the focus is just all over the place. Second, a work around worth considering if autofocus is necessary is using the higher precision center point and planning to crop if that will provide the print that one wants. Here the extra megapixels of the 1DsMKIII might help. Basically, however, what is needed is for Canon to develop more precision in the AF system and probably to use points like the center point in outer focus point. I hope this helps....Show more →
Maybe it is the AF that is not precise enough...
I don't think anymore that my particular copy of the 1Ds3 represents how the cam should be, but all my 19 AF points had issues of AF hickups (or do you spell hiccups?). Even the center one. That is: one moment the focus is fine, no change in subject and shooting conditions (even left on tripod) and the next moment focus is off (after AF confirmation) by a big margin. Totally random and unrepeatable.
Alistair Watson wrote:
Not yet mate, still going through the 1000 odd 1Ds Mark 3 shots I took yesterday and it is hard work because as much as I am starting to like the large detailed images, my computer hates them!
So, so far my upgrade cost for the 1Ds Mark 3 is likely to be £4900 for the body itself and £2500 for a top of the range Mac Pro!
As soon as I have done it, I will post, don't worry.
Yeah those 21MP files definitely make the smoke come out of your computer if it is not up to speed
Jul 07, 2008 at 10:00 AM
Steve Spencer Offline • • • • • • • Upload & Sell: On
morganb4 wrote:
My undrestanding was that the 19 user selectable focus points were all high precission.
Yes that is basically right, but it isn't quite that simple. From my reading of the white paper the center point is still more precise than the outer points. Here is a quote:
Compared with the other 18 cross-type AF points, the
center AF point's focusing line width is larger while the
fill factor for each pixel is larger (f/2.8 field of view).
Also, the number of focusing lines increases along with
the brightness (f/5.6 field of view) to attain higher
pre-cision. Gross defocus is also detected. The camera
can also detect a grossly defocused subject 1.8 times
better than with the EOS-1D Mark II N.
The 18 cross-type AF points other than the center AF point detect vertical lines at f/2.8
and horizontal lines at f/5.6. If the maximum aperture of the lens or lens/extender
combination in use is f/2.8 or larger, high-precision, cross-type focusing is possible.
So yes the outer focus points seem to have some high precision capabilities, but less so than the center point. They should be better than the system in the 1DsMKII and 1DMKII, but this may be negated somewhat by the higher magnification of the 1DsMKIII. Anyway, it is hard to know exactly what to make of all this but I thought I would raise the issue.
Ok, I did the test as you specified. I too took a blue water bottle. I shot 1.4, 2.0, 2.8, 4.0 in this order, first AF then MF, each time mis-focusing the lens in between shots. My results are about the same as yours - the AF pictures aren't as sharp as the MF ones. In my case I suspect it's because my outer AF sensors aren't "aligned" with the center one - when using the center AF point the images were perfectly in focus.
I'll get the pictures to you tomorrow offline, don't have the bandwidth today.
Was one shot under predominantly daylight, whilst the other predmonantly tungsten?
