Yes Virginia, a picture is worth a gazillion words. The following image was nicely aligned in the viewfinder:

1Ds III Misalignment

These pics all seem pretty extreme. I'd return anything that's off this much immediately.

Hi Mark, The serials ranged from 606xxx to 609xxx.

It was very obvious just by putting the camera on a tripod, aligning a shelf along the top edge of the frame and taking a pic. We saw a 4mm from from left to right on the viewing screen.

I tested the first one in the studio securing it to a stand, leveling it looking down at the floor. Place four rules on the floor so they parallel but only just visible to the four sides of the viewfinder. In the resulting image you should just see a little more of the rules on all sides but nice and parallel to the frame. This required a 1 degree correction in photoshop to achieve with the dodgy camera.

If you want to see a one degree error, make a layer in photoshop and spin it just one degree - it really is a huge error .... despite what a number of people may be suggesting on this thread.

Have you seen http://www.northlight-images.co.uk/article..._1ds3_tilt.html

I understand Canon are already doing the adjustment under warranty, but they are unlikely to admit to it wholesale as that would effect sales right now while they have a glut of orders.

1D Mk3 focus issue head in the sand denial all over again I'd wager.

The range of serial numbers is a bit concerning Carl. One wonders though why the problem varies so much between users. It does begin to sound as if Canon may be rushing them out the door before the competitors market their 20+MP offerings.

If they are already repairing them under warranty, it means they have recognized the problem. As it is appearing on at least four websites already (here, DPR, FM and Northlight) it would be difficult for them to sustain any insualrity. It looks as if the key decision facing them on this matter is whether to just keep repairing them quietly or to issue an advisory. If the noise gets loud AND widespread enough they may need to say something. But it is good to know they are fixing peoples' cameras. It would also be interesting to know exactly what the cause of the problem is.

Yes, it is tricky!

I tested only to the point where the margin of error that *I* would introduce in real world shooting is the limiting factor. I don't expect any v/f optical system (with or without a grid screen) to result in perfectly aligned images - not on the 1Ds3, or any other 35mm format camera.

In the field, I would probably work faster than when I was testing - likely resulting in images that need some straightening (but it all depends on the image, right?!).

A .75 degree rotation error *built in* to a defective 1Ds3 would result in a higher number of images that require a corrective cropping and would compound any human error introduced when aligning. Not to mention it's just simply too great an error to be acceptable on an SLR.

I still believe that the relevance of this issue, if present, is best judged by the photographer him/herself and that generalized statements aren't really valid. This is an $8000 camera and most users are probably quite astute about their photographic needs.

S.

I still can't figure exactly what the problem is. If the camera has room for adjustment in either the viewing screen mask or the position of the sensor I would of expected a test to confirm they are adjusted correctly before the bodies leave the factory.

I think I'm right in surmising that a slight mis-alignment of the mirror along it's diagonal would also cause a rotation of the image.

I guess if they are fixing it .... they can tell us! - Apparently I'll get a call from Canon UK tomorrow so I'll ask the obvious.

I'd like to make it clear that I'm not here to knock Canon, I've invested alot of hard earned cash and I'm surprised to come across such a dumb fault at this stage in a products production. It would also be nice to be treated as a valued customer with some good old up-front honesty for a change.

If you can't see it - you don't have the problem.

It is an obvious error - we are not talking about splitting hairs! (I wish we were)

A misaligned hood ornament will not affect the use of the car in any way. Incorrect viewfinders directly affect the use of a camera. And that's why people are returning the wonky items.
I rarely have to straighten horizons and I don't use tripods or grid screens either and if I do, I'd say 0.75 of a degree is surpringly significant rotate. But then when I go into people's houses all the pictures usually look wonky to me, because they are normally a little out. Some of us notice these things, some don't.
And as mentioned above, if you aren't that accurate then the misaligment may make one is less accurate to the point of wasting time in PS correcting lots of images.

Just because you think it's not a problem doesn't mean it isn't a problem. Besides for a very, very expensive camera like the 1DsIII, suggesting one bodges the screen seating or uses sticky tape to make up for a manufacturing defect are really stupid and naive suggestions.

The fact that numerous websites are reporting issues, the fact that when talking about the new Canon + Nikon in my local shop, the first thing mentioned was that all their Canons had a problem and there was even a chap in the shop who'd returned one because it had a problem, tends to indicate that they may actually be.....a problem.

Anyone who thinks this might be a trivial problem has never worked with super demanding anal art directors who really can see the difference. Anything, and I mean, anything that slows you down or forces you to recompose your image to compensate for a viewfinder/final image misalignment is completely unacceptable. Having even the slightest misalignment would drive me completely nuts. I don't want to have to guess as what my actual framing might be, especially when shooting on location, where you often don't have the luxury of either shooting tethered or using Live View. I still want a 100 percent accurate viewfinder. For the record, my 1DsMKII is extremely accurate and even with the standard focusing screen, horizons and other straight lines are right on.

It's interesting that Samy's seems to be having trouble getting cameras right now. I wonder if this issue is being addressed at the factory and might account for a supply shortage.

You put that very succinctly.

Have to say, my 1DsMKII is also extremely accurate - I repeated the test I applied to the dodgy MkIII just to clarify.... technological regress is difficult to accept!

Just collected my MkIII from Canon service (less than 24 hours turnaround, handy they are only 15 minutes away) after a prism assembly realignment and all is now as it should be.

Good to hear! Mine's all packed up and ready to get fixed after contacting Canon Service today. let's see how quick my turnaround is in the Jersey facility.

I hope it's better than the UK. The professional servce is anything but.

Nice move! ... "prism assembly realignment"... sounds like a relatively easy fix to the uninitiated (me) , I can almost feel the Canon technicians taking a deep breath prior to the arrival of hundreds of sold and unsold cameras....

I had a call from Canon UK who are not aware of a widespread problem other than one compliant (which is interesting as I've returned four myself!) Shame there is no communication between divisions on both sides of the pond... no information is good information from their perspective I suppose!

Glad you got your problem sorted. Well done Canon.

In my own testing, I took six shots on tripod. For each of them, I aligned the middle horizontal line of my Ec-D grid screen with a different horizontal line. I used the ruler tool of PS to determine the tilt. Here are the results, all CCW: 0.3, 0.34, 0.34, 0.36, 0.45, 0.49. Average = 0.38, Std Dev = 6.7

A 0.75 degree tilt is quite large, one that I would certainly see on screen and want to correct. Should my camera have exhibited such a large mis-alignment, I would have sent it immediately to CPS. As it is, I will still do so, but with less urgency.

I just don't understand how some can say that 1 or 2 degrees is within normal operating error. That's a big deviation if you have any horizontal lines in the image. Using a "cheap" bubble level, I can certainly get consistently within less than 0.5 degrees.

No offense, but from reading a large number of posts on various problems with various cameras and lenses I notice that in general photographers are very insecure in the areaas of engineering and technology. Perhaps one can't produce a nice sunset picture and know how cameras are made and work at the same time.

A fraction of a degree viewfinder/sensor misalignment is absolutely unacceptable. It is unacceptable in a fake wood box made in China bought at a home decorating store. 1 degree slope is a 0.2 inch or about 5 mm gap on the length of a letter size page (11 inches). It would cost less than a couple bucks per piece to assemble cameras with less than 0.05 degree misalignment guaranteed.

How can anyone look at a sloping horizon and need to ask someone else whether this is OK? How can anyone look at a fuzzy picture and persuade oneself it's OK because the gadget is sooooo great otherwise?

Why did they start making cameras like this?
To save money. They switched to a technology where there is no fedback or quality control - this is an "improvement" - assembly is supposed to get it it right without any check or adjustment the first time. And it works -- how well we all can see. Practically all japanese cameras are now made this way. From a $ 600 Rebel to $10,000 Mk XXLIV. How much money do they save? A layman would be surprised to know that even for a $30,000 car the manufacturers do care to save $1.50 per radio control knob. I would estimate the savings per camera (as opposed to an assembly with actual adjustment of the alignment) as a couple bucks at most. Probably much less. Does it count to them ? They are absolutely sure it does.

It is unacceptable, just as a lousy focusing accuracy is unacceptable, but it is present in practically all japanese cameras. But there is nowhere to run. Switching to Nicon won't help. Maybe Leica.

One of the complaints about the M8 (besides several people reporting going through multiple bodies that failed completely before giving up and getting a refund) is having to send lenses in for focus calibration because of discrepancies between the rangefinder focus indication and the actual focus distance...

Hi Dottore,

I would much appreciate if you could inform us how you know this is the way they manufacture an 8,000 dollar professional camera.

I know for a fact this is incorrect.

In fact they do a full QA analysis of each professional camera and 'L' lens before it leaves the factory. The problem is that there have been some training issues with the new QA team. . .

Matthew,

That ranges between humorous and insulting depending on one's point of view; but more to the point, let me ask you the same question: how do you know for fact that they do a full QA analysis on each camera and L lens? I wonder whether it isn't a sampling procedure, otherwise what seriously explains how mis-aligned viewfinder prisms would escape detection at the factory? It's not as if Canon is new to the manufacturing of these kind of csmeras.

Mark, at the end of the day, it doesn´t really matter WHICH department within the maunufacturer´s production line is at fault; what does matter is that there seems to be no worthwhile final inspection.

I worked for some years on the factory floor of R-R aero engines production; final inspection was very strong, very RESPECTED. Perhaps that´s the difference in world production ethics: the accountants now get the respect.

Rob C

I think the problem is profit... it shows itself in corner-cutting manufacturing technique.

Canon's profit is far more important than photography and the thoughts of you or I. Gone are the good old days of Leica, Zeiss, Hasselblad etc making super quality optics and cameras sometimes at a financial loss to show they are the best - and taking pride in the fact.

While professional photographers would like to believe that expensive 'Pro-end' cameras are meant for them .... the reality is that most are sold to amateurs (with more money than sense in many cases).

Canon, Nikon etc now tread a fine line between what is 'good enough' for the consuming masses and their share price.

Why did they say 'user error' was at the route of the 1D MkIII focus problem for three months? ..... to keep the cameras selling in an aggressive marketplace; you can fix them afterwards but you need to sell 'em first! (Does anyone really believe a company with millions of dollars to throw at a problem didn't identify the issue pretty quickly?)

Don't blame the monkeys, blame the accountants!

A contact I have at Canon Canada service indicates that there is a service bulletin on this issue.

Good to know - here in Toronto Canon's facility is only a 45 minute congested, hair-raising drive West on our crippled infrastructure called "the 401". Anhow its within reach. But did they point to a place on their website where this service bulletin is posted? And did they say anything about how long they keep the camera to check it and fix if necessary? Of course they have two standards of service there depending on whether or not one is CPS.

Just a joke Mark! I figure they aren't doing a full QA test including inspecting images, because a 1.5 degree offset is glaringly obvious at first glance when looking at thumbnails of horizons or architectural shots.

I do think a full QA test of all pro cameras and lenses would be a good idea though, especially one with a signed "checklist" of inspection items and a MTF test chart in the box.

I hear that CZ does the MTF sheet with their pro lenses. . .

But the monkey picture was just some light-hearted humor. Certainly it is not just Canon who is failing to do adequate QA of their products. . .

Fair enough Matthew. All this bantering though does raise a humour-neutral question about kind of QA they really do implement. I doubt they'd ever reveal it. I spoke to Canon Canada this morning, following upon guydon's post. They told me they are unaware of any service bulletin on this matter, but as the camera is under warranty they would inspect it for me with a one to two day turnaround. If they find it needs adjustment, they would want to keep it from two to three WEEKS depending on service volume at the time. If you are not CPS, that's Canon service for you. To keep in mind for those occasional "forks in the road" when one makes fundamental decisions about maintaining or switching systems. Of course it also depends on whether Nikon or the others are any better.

By reading e.g. patents in the area you learn that there are "pasive" and "active" methods, where active methods are those where actual alignment is made. Pasive methods use various smart tricks with holes, rods, markers, etc. to assembly the widget, but no check is made.

The text below is about sensor tilt, not angular misalignment. But it is quite clear that "pasive methods" are very attractive to the manufacturer.

"For high cost/low volume cameras, the CCD is "actively aligned" to the lens. This means that each individual camera is adjusted to eliminate this tilt. Currently, this is a very expensive and time-consuming process which is impractical when manufacturing low cost/high volume cameras. Consequently, most lower cost digital cameras do not actively align the CCD. Instead, they typically mount the CCD to a very flat plate (with screws) and then mount the plate to the lens assembly. By controlling the flatness of the plate and CCD package, and the perpendicularity of the lens mount to the optical centerline, the resulting tilt can often be controlled to an acceptable degree. However, not only does the CCD plate adds cost and assembly time, but it still results in greater variation of lens/CCD tilt than can be tolerated in certain applications. "


So it is, I believie, with angular alignment. it "can often be controlled to an acceptable degree". Enough said.

There definitely is a qulity control after the camera is assembled. A controller can see with naked eyes that the alignment is say, 2 degrees off, but he passes such a piece, until it is , say 5 degrees off. After the camera is assembled adjusting would cost as much as a repair, and most people will never send it for repair.

You probably should mean QC which is a part of the QA bureaucratic system. Of course they do a QC of every camera and lens. And that QC sees the same thing we all see unless they employ blind people. The problem is that companies can, and do, save money by simplifying the manufacturing, and then just asking "misalignment? what misaligment?"

Question is, is the camera build so that significantly better alignment by the service is even possible?

All of this is thoroughly unconvincing. It has no necessary bearing on or relevance to the issue of the kind of the quality control and quality assurance that Canon exercises in respect of the model 1DsMkIII. You made an unequivocal statement further back that there are no such processes for this model and I asked you to tell me how you know this. What you've said above is a pile of textbook information and inference that doesn't address the specific question.

And as for Matthew's point about whether such problems are correctable, the Canon rep who spoke to me over the phone last week told me that if they were to find a mismatch between the viewfinder image and the recorded image in my camera they can fix it. The only issue is the unacceptable amount of time over which they would keep the camera. Another poster in this thread seems to have had a satisfactory repair experience for the same issue from Canon in Australia, hence I would be inclined to believe what the Canon rep told me.

The remaining credibility issue in my mind is that the Canon employee I spoke with told me there is no service bulletin for this problem, whereas phiotoguydon further above said a Canon employee told him there is one. So there is contradictory "evidence" about whether the company has officially recognized the problem as generic enough for a service bulletin, or whether they have not. In either case, all these cameras are still under warranty and that is at least as good protection as a service bulletin, if one can leave their camera with Canon for weeks.

It is this aspect on which I would call Canon to account. While manufacturing errors and oversights are a fact of life we wish could be minimized (and at a cost they most likely can be), once it happens I think it behoves a company like Canon to ensure faster service for those people who paid over 8000 dollars and would face loss of use for several weeks. This is the aspect that "adds insult to injury" and is truly unacceptable.

My Mark III serial # 606xxx has this problem. It's very obvious - just line up anything with a straight edge using Live view and then look at what you see through the viewfinder. Or line something up through the viewfinder and then review the shot. There's no way to miss it and it doesn't require any complex testing to see it.

Anyone who has paid $8000 and actually owns this camera and has this problem would not suggest it was not a big deal. Nor would they try to suggest some kind of workaround. I did not measure the degree of tilt, but I don't care if it's off by .1 or 10 degrees, whatever it is, the fact is it is very obvious to the naked eye and therefore not acceptable.

I'll be contacting Canon to have it fixed. I understand mistakes happen so I won't get upset or hold it against them unless (or until) I experience problems with Canon's customer service. Have never had to use them before so hopefully it will be painless.

Every time there is another person highlighting this issue I haul out my camera and make another test shot. This morning's were OK. I think it is hard to test unless the problem is VERY obvious or one can set-up near lab conditions. Lenses have some distortions, the target needs to be absolutely rectangular or square, one's ability to do the line-ups and observe them needs to be spot-on etc. So testing for this problem can be a bit hit or miss. In those circumstances such people probably don't suffer from it. So my two points here: (1) based on my experience so far, the defect itself seems to be non-systematic - some cameras having it, others not, and (2) finding it can be tricky unless it is quite obvious and repeatedly so.

Mark, mine was visually very obvious if you have a line that should have been vertical or horizontal, probably 2-3 degrees off. If I had to measure it I couldn't care less. But the misalignment I had I wouldn't tolerate in a point-and-shoot.

It is inexcuasable, but having said that, if Canon is going to screw up the 1Ds3 then this is the way to do it because it's fixable and once fixed all is well. If it had been a design flaw, that would really have sucked.

An update since this morning - I contacted Canon and they had never heard of this problem before... hmmmm... But when I asked how long to have it fixed was told it had been taking about 2 days to check the problem and then 7-10 to get it fixed. Sounded to me like they had been getting them in despite the attempt to sound surprised at hearing of the issue.

I'm guessing they just hate to admit an issue that they hope might be overlooked. That would save them money and negative press so I can't blame them. But I would think most people that use a Mark III are not your typical point and shoot casual user who might not see the issue.


Mark - I would not worry about yours. It's very obvious and doesn't require any testing other than looking through the viewfinder and comparing the shot on the LCD. If your camera had the same issue it would be apparent. So rest easy you got a good one!

Sean, thanks - yes you're right - at this point I'm quite relaxed about it, but looking more generally at the situation as a whole, I can't help thinking Canon would do themselves a real PR favour by training their staff not to play dumb and just tell the truth. In the long run it pays, because then people could trust them. Each time they pull a stunt like this they erode the confidence of the very kind of customers that are important to them, while high-res full-frame is becoming an increasingly competitive market. From all the reports on numerous forums, unless it is the same handful of people reporting the same thing all over the place, this issue is more than just several cameras, but perhaps not an epidemic. Whatever, in these circumstances they should also extend themselves to assure those who need adjustments faster turnaround.

1Ds3 Serial # 603xxx, checked viewfinder alignment via Live view, a very, very slight difference. Not enough that I'm concerned or worried about getting it adjusted.

Posted this just in case someone is trying to find an affected range of serial numbers.

Canon might have this problem fix, I tested my 1Ds Mk3, today, with a EcD ruled focusing screem and it is perfect, my camera serial # 606xxx

I got a 1DsMKIII on Saturday and checked the alignment today in the studio. There was about 1/10 of a millimeter lateral shift between the viewfinder and actual image and absolutely zero rotation. Pretty damn close. Love the ergonomics, LCD, speed, interface, etc, but I have to say that in preliminary testing with very sharp lenses, there is an almost imperceptable improvement in image detail, not enough to see in any real print or reproduction. Further testing this week; we'll see...

I tested and rejected four 1Ds MkIII bodies ranging from 606xxx to 609xxx. The problem seems to be random and widespread.

People all over the world are now being told different things by Canon i.e. "no known issue", "there is a bulletin", "it has been identified as prism misalignment" etc.

My guess is that the smokescreen of misinformation will buy time while they check and tweak the unsold stocks before dispatch.

A wholesale admission of fault would be embarrassing for Canon, though I actually think a bit of honesty at an early stage would restore customer confidence!

As they say .... "Sh*t happens!" - bite the bullet and get on with it Mr Canon.

Exactly what I told them in a customer satisfaction survey they sent me by email yesterday. While I don't think I've been much affected with this current fault, one wants to be invested with a camera system backed by a company one can trust for honesty and responsiveness.

The very fact of how mute they've been despite the deluge of complaints on a number of forums demonstrates just how insular they are and what kind of corporate cultural makeover is needed at Canon. The service agent I consulted late last week told me they don't base their service policies on information from web forums. Well fine, forums would be less reliable than what they know first-hand, but the whole attitude reeks of insularity. A company more connected with their customers would respond openly to these concerns, deal with defects promptly and learn from it.

Canon has made marvelous technological breakthroughs in concepts, design and manufacturing, for which they deserve much credit. We are where we are partly due to their technological prowess, but there is more to the business than that. I think those of us reading this thread know all that - the question is whether Canon is reading us and do they?

Well I'm glad I found this thread.

Just spent the weekend in London shooting hotel interiors and aligning the camera to the doorframes, architraves or whatever (always checking that they were true - not always the case in old buildings) nearly drove me nuts.

The problem in my case (serial number 614xxx, for what it's worth) is slight, but annoying.

It should NOT be present in a camera of this expense.

D.

ps. I should add that I still had to correct a few images in ACR too. This costs time and pixels and can affect composition slightly. A pain in the butt.

Now that I've seen this thread, I'll get onto my dealer tomorrow.

Just tested mine and it looks ok thank God

geotzo,

What serial number is yours? i.e. first three digits.

Thanks.

D.

glanced at this thread and cynically thought maybe this was a case of nit picking nerdishness, till I tested mine (brand new USA model Jan 16th, 610xxx) and guess what - its out by a noticeable amount... 0.3 to 0.35 degrees.

viewfinder set up on tripod with nice clean parallel line, but the actual image opened on computer drops down on right side.

seems they're still shipping them that way from Japan.

got to send mine back to Canon now. damn it.

Wow, people with 610xxx and 614xxx still have this problem. You guys must be unlucky or everyone with the problem is posting here. Canon should have this problem fix by now. I thought they did, since my is 606xx and there was no problem with being off. Canon better fix it, there are too many people bitching about it. We all know what happen to Kodak and their DSLR. Canon Service Center should go overtime and fix this problem for owners. Paying $8000 for a body and finding out it is out of wack is not a good thing.

If you think you will ever get any over-time service from Canon, you are dreaming in Technicolor. Here in Canada Canon's service center has not admitted there is a problem, and they openly state they pay no heed to web Forums; but if you have a defective camera they will repair it under warranty provided you leave it with them for two to three weeks. I think my camera is OK and so far it seems that the image quality is excellent, but with this kind of corporate attitude and service turnaround it raises serious questions about whether to buy anything more in the future from Canon.

I've got a 614*** and I've just tested it again with the camera in portrait mode/format.

While I can't be absolutely sure, it looks OK.

The problem is that the bottom edge of the viewfinder, which becomes the right hand side edge when the camera is in portrait mode, is far from true itself. i.e. there is very slight barrel distortion because the viewfinder glass is less than perfect.

I shot a door frame and some kitchen cupboards. When I open this up in Bridge and drag the crop tool lines across the image, they align pretty perfectly with the verticals in the image.

However, ideally I'd like to be testing it with a grid screen fitted so that I can align the camera against one of the barrel-distortion free inner lines. I have one on order and I'll hold my fire until I test it with this fitted.

D.

The barrel distortion is probably from your lens. I've observed the same and I agree it interferes with evaluating alignment. I got around that issue by correcting for barrel distortion in Camera Raw's lens correction tab before checking for any off-angle of straight lines in the image once rendered in Photoshop.

Hi Mark,

You may be right, but Canon's 50mm 2.5 Macro is considered to have zip distortion.

The viewfinder, on the other hand, is another matter.

D.

I posted this on DPR as well, but figured some people here might be able to add their input...(summary is - my viewfinder was tilted, sent to Canon, my viewfinder is still tilted)...

I received my camera with a serial number 606xxx, and after shooting a number of landscapes, noticed the tilt in the viewfinder.

I took these shots to show the variation - there's a 0.8deg tilt:


I called Canon Irvine, they told me to bring the camera in. I dropped it off, and after 2.5 weeks (with no updates!), finally saw 'Completed' on the repair status page. I picked it up a couple days ago, and shot a new test image (lined up like the last one, but with grid paper and center crop. If people really want a through-the-viewfinder shot, I can do that too...but I was lazy at the time):



Canon didn't fix the problem! Instead of 0.8deg, it's between 0.5deg and 0.6deg.

What do I do now? It wasn't a particularly good feeling to not have the camera for over two weeks and miss out on shoots, and it doesn't appear as if Canon Irvine can even do the repair accurately. Is there anyone I can contact?

That's what I suspected: since these camera viewfinders and/or sensors are assembled using "passive" methods there is no possibility of positive adjustment. The repair shop can try to re-assembly it, but the result is more or less random.

If there was an actual tilt adjustment, someone would finally find out how to do it at home. I don't care to prove it to anyone, you can believe it or not, but the evidence indicates that the angular alignment in the trio camera baseline-vewfinder-sensor is not adjustable. Maybe we should admire the wonderful technology that allows less than 2 degrees accuracy without adjustment in most cases. Wow.....

I don't see any remedy to that situation, except screaming and trying to shame and ridicule all camera makers whose lemon cameras produce so obviously sloping horizons.