p.4 #1 · 1Ds3 AF: learning to understand how it works
Andrew Gough wrote:
I could not agree more.
I also wonder if flare is part of the issue at f/2.0 with the bright background.
DaanB.
I would suggest re-doing the test on a brick wall of equal distance that is under overcast light. Bright sunlight would give a very contrasty image for the AF to work with.
I did the same shoot both under sunny and overcast conditions. I didn't see any changes. Not even when the sun shined full on the side of the chimney facing me. About flare, I took the shots in the OP with the sun coming from behind (sun was behind clouds also). There is also no flare present in the pictures.
I think it is just that the small AF sensor in combination with the "muddy" contrast of the bricks at that distance (about 25 meters) that is fooling the AF. In other words, it just can't read enough contrast of the bricks at 25 meters for a proper lock. Not enough details to lock on.
I could try the same shot at different distances, see if that would make any difference. I know that when shooting at 1-5 meters I don't have AF locking problems when aiming over/on top of a object. The question is, at what distance do these lock-on problems start to appear when aiming over/on top of the subject.
p.4 #2 · 1Ds3 AF: learning to understand how it works
Markuson wrote:
Note the strange ghosting effect off of the left side of the trunk edge (yet another issue Canon has not admitted to, but that has been heavily documented on this and other forums, as well as by numerous respected reviewers).
By the way... The ghosting problem seems to be random, and can happen at ANY time for no reason. -Even shot to identical, tripod-mounted shot.
I welcome comments.
The result is undeniable.
That doesn't look right
I think your 1D3 is broken... Either a hardware or software malfunction. I had that same kind of weird AF shit with my previous 1Ds3. Only with different lenses (I don't own a 500L). Nothing helped. I even got it and my lenses calibrated by Canon Service three times. The weird, erratic and random mis-focus remained. In the end they replaced it with a new 1Ds3 that functions fine.
p.4 #3 · 1Ds3 AF: learning to understand how it works
Daan B wrote:
I did the same shoot both under sunny and overcast conditions. I didn't see any changes. Not even when the sun shined full on the side of the chimney facing me. About flare, I took the shots in the OP with the sun coming from behind (sun was behind clouds also). There is also no flare present in the pictures.
I think it is just that the small AF sensor in combination with the "muddy" contrast of the bricks at that distance (about 25 meters) that is fooling the AF. In other words, it just can't read enough contrast of the bricks at 25 meters for a proper lock. Not enough details to lock on.
I could try the same shot at different distances, see if that would make any difference. I know that when shooting at 1-5 meters I don't have AF locking problems when aiming over/on top of a object. The question is, at what distance do these lock-on problems start to appear when aiming over/on top of the subject....Show more →
If it does not focus with the chimney under full sun, then I would say that your camera is defective, and you should return it to Canon.
Markuson, you have attempted to hijack this thread with your AF problems on a different camera. While I sympathize and agree with you, you really should start another thread to voice your complaints.
Edited on Aug 16, 2008 at 07:35 AM
Aug 16, 2008 at 07:32 AM
brainiac Offline [X]
p.4 #4 · 1Ds3 AF: learning to understand how it works
Hrow wrote:
The theory behind the AF may be the same but many of the components are and have to be different simply to account for the different sensor sizes.
No. Canon economises by using the same af components in both pro bodies. That's why the 1D3 AF points take up most of the frame, when the 1Ds3 points only occupy the center. It's mostly the same AF system.
Rob Galbraith's latest report suggests that any differences in performance between the two probably result from frame rate and software tuning.
I use a 1Ds3 mostly in manual focus with a Canon manual focus screen. I have never trusted AF, so the quirks/failings of this system don't bother me too much. However, I don't shoot sport. When I need maximum accuracy and have time, I use liveview at 10x magnification.
The 1Ds3 is a great camera and there are other reasons to buy it than AF reliability. The extent to which AF perfection matters will probably depend on the type of photography you specialise in.
p.4 #5 · 1Ds3 AF: learning to understand how it works
Andrew Gough wrote:
If it does not focus with the chimney under full sun, then I would say that your camera is defective, and you should return it to Canon.
There is simply not enough distinguishable detail to make a proper contrast reading at those distances. The area covered by the AF sensor when focusing on the middle of the chimney is too uniform in its appearance (only "muddy" contrast of bricks and morter). When aiming for the edges, all is in focus. If the cam was defect, it shouldn't be doing that either. I think you are on the wrong track here
p.4 #7 · 1Ds3 AF: learning to understand how it works
ghozer wrote:
There's no doubt some work fine. There's also no doubt some don't. For $4500 that's pathetic.
And that's why Markuson is so angry. (rightfully so)
I do believe that companies should stand behind their products----Markuson deserves either a repair or another camera body, so he too can enjoy what I believe to be the best DSLR to date. (Okay---so he needs the 1DS Mark III...)
For now, I'd suggest resetting the lens microadjustment features to -0- on the camera body---and starting over with the Cfn settings..... If indeed it is a QC issue (as I believe) setting the microadjustment feature may hinder and not help the situation. It is the lack of the AF system to consistently lock focus on a fixed point. By using the microadjustment feature---it is difficult to test the AF system. My suggestion is to use the microadjustment feature only if you are confident in the consistency of your AF system and you feel sharpness could be better with a particular lens.
Fingers-crossed for you, Markuson-----because I want to be really confident the next time Canon comes around wanting me to empty my wallet of $8K.....
p.4 #8 · 1Ds3 AF: learning to understand how it works
Kagetsu wrote:
The C.FnIII settings are:
5 = 1
6 = 4
13 = 1
All others = 0
2, 16, 17 = 0.
Function 5 is the one that made all the difference though, as the others are effectively superficial.
Just gettings results now, and will post shortly.
Thanks
So basically you tell the cam not to re-focus. Did you use AI Servo? Interesting to see if that will benefit low light AF. I will do some testing myself. Please post the findings on your results
p.4 #9 · 1Ds3 AF: learning to understand how it works
Btw, I don't have the same Cfn settings that you ----but my 1Ds Mark III performs like a stunner in dark conditions. It is able to lock focus in places (and quickly!) that I would never even dream of taking a photo---and where image quality would certainly reflect that. In otherwords, the AF in dark situations performs beyond the capabilities of the camera to produce a suitable image.
All images on 1Ds III, all exif intact. Note that all shots on the stip are taken with the 70-200 IS f/2.8L. All shots shown are the first shot using AI autofocus, at ISO 1600 in AV mode.
Focus point is centre point.
First lots of shots are for other people, but you're welcome to see. Even at relatively low resolution that these are (crops will come later) the track shots are clearly in focus with the centre point. Lighting while the photo's make it appear that it is good, it's far from it. The bird flying on the track is spot on with focus and is a one off shot (first and only shot fired). Considering the effort that would have gone into the AI shot, it's turned out remarkably well.
EDIT: Before I forget, I was in a very I couldn't care less mode last night, and was only there to get one photo (the cruddy looking white wagon) for a friend. ^_^'
Don't expect perfect photo's, no post processing has occured.
p.4 #11 · 1Ds3 AF: learning to understand how it works
And you, amaze me.
I've had an 'S' for almost a year and use it primarily for birds in absolutely the most trying conditions. It's a phenomenal camera
90% of this talk is coming from people who've never owned one, and the other ten percent is coming from people who obviously can't operate one.
Markuson wrote:
I am absolutely and continually ASTONISHED at the level of dismissive attitudes I see from photographers regarding this absolutely UNACCEPTABLE performance in a top-level camera.
Some seem to think "top-tier" means you should be some sort of bizarra AF expert before you should expect it to work. The OPPOSITE should be true.
I've used numerous 1 Series cameras, and NEVER has AF been a CHORE.
NEVER has AF been difficult.
NEVER...until I bought the 1D3.
It is SCREWED UP, people!!
-And no amount of excuses...explanations...and outright blindness will change this fact!
You guys just AMAZE ME!!!!!
Sheesh.
"AF is always difficult at infinity" What the HECK??
What planet are you from!??
That has NEVER been a problem unless pointing at a detail-free blue sky!
What the heck is happening to photog mentality these days that so many will come up with these absolutely stunning and ridiculous excuses for a $4500 MESS??
p.4 #12 · 1Ds3 AF: learning to understand how it works
Markuson wrote:
Where did I claim people didn't know what they are doing??
I am specifically reacting against that very thing being claimed about me! Not the reverse, as you imply.
This is very simple:
-If I put my 1D3 and 500 f4 L lens on a heavy-duty carbon fiber tripod and Wimberly head...in sunlight...and point it at an easy-contrast tree-trunk...with properly adjusted microadjustment...updated firmware, and all of Canon's supposed "fixes"...in ONE-SHOT mode...(did I mention Wimberly-mounted and rock-solid??)...it should NOT look like this:
Note the strange ghosting effect off of the left side of the trunk edge (yet another issue Canon has not admitted to, but that has been heavily documented on this and other forums, as well as by numerous respected reviewers).
By the way... The ghosting problem seems to be random, and can happen at ANY time for no reason. -Even shot to identical, tripod-mounted shot.
We have what looks to be a similar sized tree in our backyard which has wonderful contrast on its bark during golden hour. I've used that tree of 1000's of test pictures - and 100's and 100's of times the cameras have back focused on the edge of the tree, not on the center of trunk. Sometimes they get it right, but most often not. I've tested the 1D2, 1Ds, 1Ds2, 1Ds3, 40D, 10D and just about every Canon mid range tele & tele zoom on that stupid tree. And many times camera bodies and lenses went off to Canon (Irvine) for calibration. More or less the same results on their return. These same cameras performed excellent on real world shooting, so using tree trunks as test targets isn't the best subject. I'm not saying that you don't know how to use your gear, just that the test scenario isn't very reliable.
p.4 #13 · 1Ds3 AF: learning to understand how it works
Johnny Bravo wrote:
90% of this talk is coming from people who've never owned one, and the other ten percent is coming from people who obviously can't operate one.
Provocative and probably untrue. I'm satisfied with the AF of my 1Ds3 but some of the critics may well have defective cameras. You can't know that they don't.
p.4 #14 · 1Ds3 AF: learning to understand how it works
John Black wrote:
"...These same cameras performed excellent on real world shooting, so using tree trunks as test targets isn't the best subject. I'm not saying that you don't know how to use your gear, just that the test scenario isn't very reliable."
That's all good and well, unless of course your subject happens to be a tree trunk. Unfortunately, many 'real-world' subjects happen to be similar to tree trunks, which my 1Ds, 30D, and 40D can focus reliably upon. I'm not sure why a 'more advanced' AF system would require an edge to get accurate AF.
Johnny Bravo wrote:
90% of this talk is coming from people who've never owned one, and the other ten percent is coming from people who obviously can't operate one.
Nothing like stirring the pot, just for the sake of sitting back and watching what happens.
p.4 #15 · 1Ds3 AF: learning to understand how it works
Johnny Bravo, judging from your response, you must be a skilled user of your Mk III. Since I'm interested in responses to the question asked by the OP - advice on using the Mk III for those coming from lower tech AF (esp earlier 1D AF tech) - would you consider offering your insights for those of us who don't possess your level of skill and have had trouble with Mk III cameras?
Edited by Mike Tuomey on Aug 17, 2008 at 01:38 PM GMT
p.4 #16 · 1Ds3 AF: learning to understand how it works
Jeff wrote:
I'm not sure why a 'more advanced' AF system would require an edge to get accurate AF.
Probably because it has smaller AF sensors and reads contrast differently in comparison to xxD and xxxD series AF. Both smaller and bigger AF sensors each have their own advantages and disadvantages. In a way it is like comparing apples to oranges.
Anyway, since it is obvious that placing the AF sensor on top of the tree doesn't give a in-focus shot with a mkIII AF, it is better to aim for the edges. Wether we like it or not. This is how it seems to work the best (with proper working mkIII's)
p.4 #17 · 1Ds3 AF: learning to understand how it works
Daan B wrote:
FWIW
I redid the shots I posted in the OP using the following methods:
1) focus in the middle of the chimney using One Shot with the following C.Fn settings:
C.FnIII-8-0 (disable AF point expansion): still OOF
C.FnIII-8-1 (enable AF point expansion L/R): still OOF
C.FnIII-8-2 (enable AF point selection area): still OOF
2) focus in the middle of the chimney using AI Servo with the following C.Fn settings:
C.FnIII-8-0 (disable AF point expansion): still OOF
C.FnIII-8-1 (enable AF point expansion L/R): still OOF
C.FnIII-8-2 (enable AF point selection area): still OOF
3) focus on the edge of the chimney using One Shot with the following C.Fn settings:
C.FnIII-8-0 (disable AF point expansion): IN FOCUS
C.FnIII-8-1 (enable AF point expansion L/R): Some in focus / some oof
C.FnIII-8-2 (enable AF point selection area): still OOF
4) focus on the edge of the chimney using AI Servo with the following C.Fn settings:
C.FnIII-8-0 (disable AF point expansion): IN FOCUS
C.FnIII-8-1 (enable AF point expansion L/R): IN FOCUS
C.FnIII-8-2 (enable AF point selection area): still OOF
At shorter distances (like 10 meters/ 30 feet and less) the AF is doing fine when the AF point is placed over/on top of the subject. With longer shooting distances (infinity) the only reliable way to get good focus is to aim for edges, rather than placing an AF point over/on top of the subject. The effect of enabling AF point expansion when focusing at longer distances with both aiming tactics (placing the AF point over/on top of the subject and placing the AF point over the edge of a subject) and using both One Shot and AI Servo is nihil (except for enabling L/R expansion in AI Servo - but this seems to have little benefit over disabling AF point expansion). YMMV ...Show more →
Helpful post Daan.
I've long been used to focusing on objects that may not be my main subject or focal point. I "almost" always look for contrast first, understand my DOF vis-a-vis my aperture second, and then choose where to focus the camera. I've done this for many years, whenever possible that is, regardless of whether I'm using a 1 series or something less (especially with manual focus lenses because it's easier).
I think the best result of the Mark III fiasco (that's both the 1D and 1Ds) could be Canon's willingness to admit they pooched it. I've given up on the Mark III series and don't think I'll buy another (at least for a while), but since the IQ (when focused) is so good, I'd be willing to try again in the next "version" that will probably come out in 2009.
p.4 #18 · 1Ds3 AF: learning to understand how it works
Daan B wrote:
Probably because it has smaller AF sensors and reads contrast differently in comparison to xxD and xxxD series AF. Both smaller and bigger AF sensors each have their own advantages and disadvantages. In a way it is like comparing apples to oranges.
Anyway, since it is obvious that placing the AF sensor on top of the tree doesn't give a in-focus shot with a mkIII AF, it is better to aim for the edges. Wether we like it or not. This is how it seems to work the best (with proper working mkIII's)
Where has Canon detailed the information of the 1D MkIII's AF sensors being smaller than those in the 5D? I don't remember seeing that in the White Paper. I think the 40D White Paper had a statement about reducing pixel pitch in relation to the 30D, IIRC.
What I do remember is the fact that the MkIII's AF sensor(s) cover a significantly larger portion of the viewfinder than the area that the little red square presents. In other words, the little red squares are not accurate, as many of us have found with many cameras. How the 5D fares here, I have no idea. I'd love to know just how large an area beyond the red square it is, and if it impinges upon the area of other adjacent (assist) AF points.
I'm no engineer, but it sure sounds like there were a lot of ways to screw this new AF system up, and apparently Canon secretly implemented one of them.
p.4 #19 · 1Ds3 AF: learning to understand how it works
Jeff wrote:
I'm no engineer, but it sure sounds like there were a lot of ways to screw this new AF system up, and apparently Canon secretly implemented one of them.
Secretly or unwittingly? Perhaps they implemented several ways..
p.4 #20 · 1Ds3 AF: learning to understand how it works
Unwittingly, I'm sure, and from the sequence of events, I'd guess more than one, only one of which was 'easily' corrected. Once corrected, only then could it be shown that something else was going on, under limited shooting conditions or circumstances (I'm guessing deeper within the design of the AF system).
