I noticed a couple of posts on DP preview and Northlight about problems people are noticing on their Canon 1ds Mark 3 that there is from a degree to 5 degrees tilt between what they are seeing on the viewfinder and what they get from the sensor or live view. The horizon seems to tilt and a correction ccw needs to be taken to line it up.

Being a portrait photographer, I don't think I would have noticed this and perhaps it was present in other DSLRs that I've had. On top of that I never print full frame 35mm anyway but prefer the 8x10 crop.

Has anyone else noticed this on their new or even their older DSLRs.


Michael

If you don't have the grid line focusing screen, getting horizons perfectly level can be hard. I've never had any alignment issues with 3 different Canon DSLRs.

Ditto. I spent 2 days walking around Hampton Court Palace over Christmas week. While I do have a few that need tweaking (all were developed with DPP which doesn't have a leveling option) most were fine.

mike

http://www.pbase.com/mike_broomfield/hampton_ct_and_molesey

I saw those posts too but I have not noticed anything out of line on my 1DsIII. I suppose if you look hard enough at any camera, you can find some kind of flaw, however minor.

I haven't seen any problems yet, in fact I'm still very impressed with the thing ..

The grid on live view is handy for tripod stuff.
(not that I ever have time for tripods, but if I did ! )

I heard of that problem with the Mark III too.

I have a Mark II and do a lot of architecture with it. If I level with the grid screen or a bubble level (very exact one, not the junk you stick to the hotshoe) there is a misalignment of the sensor of about 1.5 degree. That sounds not to be much. But for my architecture work it is clearly visible and I have to align all architecture shots.

So this is not new to me, but 5 degree - that is a joke.

Best,
Johannes

I have also experienced sensor misalignment problems with 1Ds Mk3's.

Being a very satisfied 1DsMk2 user for three years I ordered my 1Ds Mk3 about six months ago. The body (serial 609xxx) arrived six days ago. On testing I discovered it had an obvious clockwise misalignment problem between the viewfinder and the sensor.

The error was about 1 degree (that's the rotation required in Photoshop to put it right). It doesn't sound like much but just imagine carefully lining a horizontal line along the top edge of the frame in the viewfinder and taking a photograph. On the beautiful new rear screen the horizontal line drops about 4mm from left to right - a huge discrepancy.

Also, because of the degree of the error part of the image is turned out of frame completely! - hardly a 100% viewfinder!!

I returned to my dealer here in the UK and demonstrated the problem. He looked both surprised and bemused but instantly offered me a replacement which we tested there and then in the shop. Guess what - same problem!

My dealer offered to contact Canon to discuss the issue and find out when they would be receiving more units. I have known my dealer for many years and was happy to leave the camera (and the thousands of pounds of dept on my credit card) for a few days while he sought an answer.

He has since received two more bodies which we have tested together. They also have the same problem. All FOUR bodies have now gone back to Canon as 'dead on arrival'.

Is it just a batch? ..... serials ranged from 606xxx to 609xxx (yes, the later cameras to arrive did have lower numbers) - that seems like a whole lot of cameras to me!

I also contacted Canon who told me I'd have a response in 3 days.... no response.

I am not a camera technician, but suspect one of three things could be at fault. Either the sensor isn't secured in the correct position, the viewfinder mask is not in the correct position, or the mirror is not aligned correctly across it's diagonal. Whatever, it's a real dumb mistake not related to the cutting edge technology that has been squeezed into this otherwise stunning piece of engineering.

I understand Canon are accepting cameras to be adjusted under warranty... but do you really want someone taking your brand new almost £6000 camera to bits before you've taken a single snap?!! - I don't.

I've noticed that a number of people have dismissed the error as 'nit picking' but as a professional photographer who relies on a grid screen in a viewfinder to line up compositions on a daily basis... the fault makes an otherwise great tool useless.

My dealer has returned my money.... for the time being.

P.S. I've been looking at the LL website / forums here for a few years and have gleaned some great info (and purchased some excellent videos) This is my first post so I'd just like to apologize for the rather negative content. Anyway - hello to all!

Sorry for the OT question - What kind of level is your "very exact one" - can you post a link? The only ones I've found with a method for attaching to a camera are the hotshoe ones, which are hit or miss in accuracy.

Thanks!

I took mine out for some landscapes last weekend and hadn't noticed, but I just checked mine with my 50" TV and a test pattern for more reference, though just aligning the top would give you a clue. Noticeable. I wonder if it's a side-effect of the sensor cleaning vibration mechanism which makes it harder to align precisely at assembly time.

But here's the workaround, at least until Canon decides what to do, for precision shots: use Live View to frame your shot - perfect alignment guaranteed every time . Kind of a bit of a fly in the ointment. So far I love the camera, both operationally and the resulting images I've been getting, and frankly I can't really tell in the tree-laden images I shot or the wider landscapes with lots of horizon. Architectural shots would probably be a different story though.

I can see the next Rob Galbraith saga coming up ...

Has anyone with this "problem" bothered to check that their focusing screen is aligned/seated properly? Given that the 1-series' screens can be removed and replaced, it would seem that reseating the screen would be a quick and effective fix to this problem without the hassle of returning/replacing the camera. A bit of tape applied to the right spot on the edge of the screen would easily fix this issue permanently; returning the camera over such a trivial thing seems a bit of an overreaction.

This issue may seem trivial to an amateur using the camera for his/her holiday snaps, but if you are earning a living from it's use then it's a different game altogether. A camera of the type is supposed to save time and hassle - not give the photographer something else to think about.

With reference to your "easy fix" the position of the removable screen has no effect on the position of the mask attached to the lower face of the prism. It's the relationship between this mask, the mirror and the sensor that is at fault.

There is no getting away from the fact that a £6000 camera intended for professional use should produce images as per the composition in the viewfinder - out of the box!

Before joining the Army, I did earn my living with my cameras. My experience is that "eyeballing" horizons/verticals is tough to do more accurately than +/- 2 degrees or so, especially if they are near the center of the frame and your reference is is the edge of the frame. If you're serious about perfect horizon/vertical alignment via the viewfinder, you should be using the grid focusing screen, in which case the tape trick I mentioned in my previous post will work just fine, if there are any remaining alignment issues switching screens doesn't solve.

In any event, using Live View tethered is an even better option, and renders the issue moot.

Yes I am serious about perfect horizon/vertical alignment (not just of horizons) via the viewfinder, yes I do use a grid screen all the time. I also have the camera on either a studio stand or tripod 99% of the time - the operation of the camera should improve my workflow....not make it go backwards.

No, your bit of tape fix method will not work - I don't think you are understanding either the problem or the fact that an investment of this amount of money should result in a 100% correlation between viewfinder and sensor - not sticking bits of tape on the inside of a £6000 camera body to make the image 'appear' correct.

Having to correct each image by rotation on Photoshop will result in interpolation of pixels, whats the point of upgrading from a smaller pixel count just to shove the new pixels through an additional degrading hoop?

No, live view is not a solution rendering the issue moot. It is a pain in the butt that shouldn't be necessary. All I'm asking for is a capture of what I see in the viewfinder. I can see you feel I'm being unreasonable. So maybe we should move on.

I'm afraid I have to agree with Carl. I suggested Liveview as a stop-gap solution, not a permanent one, while the problem gets resolved. It is a bit of pain in the butt, and chews up the battery and gets the sensor hot. While it's probably a handy thing to have on occasion, I'm not crazy about using it as my main shooting mode. I really do like the new Info LCD mode though, which I was planning on having on all the time.

Knowing Canon's mute tendency until the very last moment when dealing with this kind of thing, let's see how long this is going to take ...

The one in the image. Sorry the website is only in german:
http://www.manufactum.de/Produkt/0/761690/...iff=wasserwaage

[attachmentid=4601]

You hold it to the front end of the lens barrel first for vertical alignment. Then you put it on the hot shoe for horizontal alignment, measure this a second time with the level turned 180° to find out whether the hot shoe is not in level.

Best,
Johannes

Apparently a 24MP full-frame Sony Alpha is imminently going to be announced at PMA with built-in sensor-based antishake for $3000-$4000 or so. A Nikon D3x is also expected albeit probably in the $5000 range, and Nikon offers the incomparable 14-24/2.8 which absolutely thrashes every Canon wide angle lens, prime or zoom, in its range.

Perhaps some potential buyers of the 1DsIII will consider whether it is worth paying almost ten grand for a body with slapdash assembly standards and QC, using subpar wide angle glass and a defective autofocus system.

It's time for Canon to step up to the plate and stop acting like they own the high end marketplace, permanently, and can charge any price for their products.

Something like this happened to a Pentax Super Program I owned once. I had bought it second hand and discovered the problem after photographing artwork, which I had been very careful to frame properly using a tripod. I thought I might have introduced a bias because of bad technique (hitting the shutter button too hard maybe; couldn't remember if I had used the cable release) but when I tried again using a different camera, the tilt went away. I took it into a repair shop and the tech refused to believe me. He put the camera into a device, exposed a couple of frames and then showed me that the exposed image was perfectly parallel with the film edges. I tried to tell him that the prism (or penta-mirror) was out of alignment, not the film transport, but he refused to look at it more closely and blamed it on user error. I figure a previous owner might have knocked the prism (or mirror, or focus screen) out of alignment.

I feel the same as some of the earlier posters about the issue in the current camera. If the problem is real, assuming it is based on the presented evidence, for that kind of money I'd be upset and would not be satisfied with quick fixes to compensate for the tilt. Buying new equipment for top dollar is what you do when you want to avoid this kind of problem. Whether I buy a Kia or a Porsche, I expect it to be in alignment when I take delivery; I'd be a lot more upset with the Porsche than the Kia if they were not, however, but I wouldn't be happy either way.

I do agree that Canon have rested on their laurels for too long and the result seems to be showing it's presence in the 'professional' end of their range with this and the 1D Mk3.

I don't think that four faulty 1Ds Mk3 cameras in a row is just bad luck on my part, so I will be looking very carefully at the next one I receive. I do like the camera, the results and handling - as I did with my Mk2 so I have no intention of changing to Nikon (although I do still use F3 bodies when I'm traveling).

I have little interest in high ISO performance so with it's lower pixel count than my 1DsMk2 the D3 would be a step backwards in it's present form.

If I have any more doubt's or problems with the 1Ds Mk3 I'll stick with me Mk2 and medium format back for the near future.

I called Canon again today as they failed to respond within the promised three days. Apparently they have no knowledge of the issue (surprise) but it's been given a high priority and I should here from someone on Monday..... don't hold your breath.

Well the D3x is rumored to be announced at the end of the month with 24MP.

We'll see if it happens or not. Canon can certainly use the competition!

I used to be a surveyor. There's no way in the world that any small bubble level device is sensitive enough to allow you to level a camera with respect to the local horizon to any greater tolerance than about plus or minus 5 degrees. They just aren't that precise. The really precise bubble levels, like the ones used on high order survey instruments, are so sensitive that if you shine a flashlight on one end (at night of course), you can see the bubble start to move as the glass vial is heated differentially by the light. Those are good to under a degree (or maybe a lot more, depending on the instrument). Also, in that kind of work, rather than just center the bubble, we read it's actual location against numbered marks on the vial, at several different pointing positions.

Knowing that, if we're interested in absolute alignment, we need to align the image to the local level plane. Levelling the tripod or the camera body is insufficient. If there's an alignment problem between the image (either the digital sensor or the film frame window) and the viewfinder, then you've got to either get the misalignment corrected, or in the case of a digital camera, use live view to match the image itself with the horizon. Thus Jonathan's suggestion. You don't have to shoot with live view, just use it to verify your setup, and then shoot using whatever other method you prefer. Of course, that won't do you much good for handheld work, but if you're trying to shoot architecture or something of that nature handheld, you've got all kinds of other problems.

When I'm shooting panoramas for QuicktimeVR, I use an ancient Canon point and shoot, since all the extra data from a DSLR would be more storage trouble than I need for my low-res result. That particular camera has about a 5 degree sensor/camera level discrepancy (I've heard anectdotal evidence that many or most of the same model are similar; probably an engineering/packaging design decision). I use a bubble level to set the tripod. Then I turn on the live view and rotate the camera around the vertical axis to check alignment. I use the tripod head to adjust the horizontal angle of the camera until I'm satisfied that the images will be level. Then I shoot. Piece of cake, and neatly sidesteps all the issues associated with tripod and camera leveling.

Jonathan's advice works. If you align correctly the line of the grids then you will not need to rotate the image in PS.

The problem that his fix does solve is the crop since the wrong positioning of the mask of the camera will result in the viewfinder not being in fact 100%. In other words, you will shoot a correctly oriented image, but you will not see the full area of the image. Depending on the mask's misalignement, this could make you waste as many as 1 or 2 MP or sensor real estate.

As a side comment, the orientation sensor of the D3 is key, since it enables the photograph - even on the fly - to align his images much better, which reduces the need to crop, and therefore makes you win as much as 10% pixel counts compared to images taken with a camera without this feature. Thanks to it a 20MP D3x would practically have the same pixel count as a 22MP Mamiya ZD.

Cheers,
Bernard

I don't think this will happen.

Nikon's D3x accouncement schedule is a balancing act between letting too many people buy a 1ds3 and preventing too many people from buying a D3 first.

IMHO, we are still way too close to the D3 introduction to make it interesting for Nikon to introduce the D3x now even if it were ready to ship. What they want is people to buy both, starting with a D3.

The only thing that might force Nikon to act quick would be the announcement of the Sony 24MP, but this is still only rumours at this point of time.

We will see.

Cheers,
Bernard

The autofocus issue with the 1D-MkIII was fixed before the 1Ds-MkIII was even released, and there's been zero evidence so far that the AF issue has cropped up in the 1Ds-MkIII. Your snarky comment about autofocus has no factual basis whatsoever.

The tape trick I mentioned will work to micro-align the grid focusing screen in the viewfinder if simply removing and re-seating the screen doesn't fix the problem. If you're not using a grid screen, then you aren't going to get accuracy better than +-2 degrees or so because eyeballing simply isn't any more accurate than that. If you use a bubble level attached to the camera, you can get within about +- 1 degree if you can adjust the level's attachment to the camera so that the camera and the level agree on what level is. Otherwise you are consistently going to be off in most cases. But even assuming you do all of that (and you probably haven't), user error can still creep in, like pressing the shutter release by hand instead of using a remote release. That can introduce an error of a couple of degrees even if the camera was leveled perfectly during focus & composition. And if you're shooting handheld and complaining about a degree or two of "misalignment"...

Canon's wide lenses (with the possible exception of the new 14mm prime) are the weakest part of Canon's lens lineup. Congratulations, one out of three ain't bad.

Just to add numbers to this post as I haven't found a viable work around, but I have a 1dsm2 that has this misalignment problem and it is a pain. Whether I'm shooting in a very tight situation where cropping just isn't afforded or another step in the workflow I really don't believe an $8000 instrument should possess this type of problem. (you want me to place tape where? taking away viewfinder real estate, and if it melts? - due to a high temperature zone?)

I never remember having a misalignment problem such as this on any of the view cameras I've owned nor having any problem leveling any of the 1Ns with standard bubble levels. Because the camera is labeled for studio I have always thought it would be nice if Canon would create flat surfaces on various regions of the body designed so the user could use a more precise leveling device than just the hotshoe. I would rather create a cool picture than having a camera designed with eloquent curves.

why would nikon offer a 24 mp ff camera in the same price range as their 12mp ff offering? the fact is that every company prices items where they believe the market is. one could make the case that inspite of it's robust features , that $5000 for a 12mp ff nikon might be pricey given the fact that there is a 12mp ff canon currently selling for nearly $3000 less. the market however, clearly says that plenty of people do not feel the nikon is overpriced and the long line of people waiting to take delivery of the 1ds3 suggest that there is indeed a market for the canon flagship. for those who do have issues with how canon and nikon have priced their flagships, they can elect not to buy them and/or look at other options that are more in line with their pocket books and /or needs. if we think the price is too high, no one is putting a gun to any of our heads and forcing us to buy the item.

while sony may announce a 24 mp ff camera, i believe their lens lineup falls far short of the offerings from nikon and canon, which should be a consideration for anyone considering a ff alpha. given the fact that sony has priced their alpha bodies to date at the same level as "comparable" offerings from other maufacturers, why would they price a 24 mp ff body so much lower that nikon and canon? are they not profit driven too?
It also seems to be a stretch to "round up " the $7900 price tag of the 1ds3 to "almost ten grand." you also might want to consider that not every shooter wants or needs an ultra wide. there have also been plenty of photographers who have made a very decent living using that " subpar wide angle glass". while people have complained about canon wides, and maintained nikon had better wides, if you take the kodak n model out of the equation, until the d3, their has been no nikon ff dslr that 'tested" any of their lenses the way the 1ds cameras have tested the canon lense line.
the botton line is that the recent releases (lenses and bodies), should make most of the camera companies assess or reassess their strategies---and hopefully adapt where necessary and appropriate. if they do that, every photographer from the season pro to weekend soccer parent will benefit.

i guess i should disclose that i am one of those people shooting with a body "with slapdash assembly standards and QC, using subpar wide angle glass and a defective autofocus system."
regards

Because the Sony with the same chip will be retailing for $3000 to $4000 with built-in antishake.



I never said the D3 was a bargain.

The Alpha 900 at 24MP and $3000-4000 will be though, and it will force the other players to adjust pricing.



If Sony delivers the 24MP for $3000 (that is the rumor, so low I find it hard to believe, but it comes from someone who has been spot-on with other Sony rumors in the past) they will sell way more of them than Canon sells 1DsIIIs. Canon will be forced to respond.



How many lenses do you think a landscape photographer needs? Come on! Tons of enthusiasts will buy a 24MP sony at $3000 who would never pay $8000 for a camera.



I don't know, but the rumor from (so far) reliable sources is $3000. That is so low that some people are suggesting closer to $4000, but the most reliable source says $3000.

Apparently Sony is trying to compete with the 5DII.

The rumor could be wrong, but that is what the rumor says.



It's not much of a stretch, $8000 and $10,000 are pretty comparable in terms of how many people are willing to spend that much on a rapidly depreciating camera body (answer: not many).

There are droves of enthusiasts willing to spend $3000 on a camera body that is worth it, witness the massive success of the 5D in the landscape market where it has outsold the 1DsII 5 to 1 or more.


Anyway, if the rumor holds true and Sony releases the Alpha 900 with 24MP at $3000 (and I hope the 900 also has the Sony live twist-and-flip LCD leaked in pictures of the Alpha 300 back) I think this will be a serious game-changer for the high-end dSLR market, the way the original digital rebel at $999 was a serious game-changer for the dSLR market as a whole. The camera makers have been charging far too much money to people who need maximum resolution, and bringing the price down to a reasonable level is good for everyone.

Rob Galbraith and many others don't seem to think it's fixed yet. And there is no reason to think the 1DsIII is any better than the "blue dot" or repaired 1DIII cameras.

But to be fair, for a studio camera, I'm sure the 1DsIII autofocus is more than adequate, so that was admittedly a cheap shot (although one that Canon deserves).



Yes, I want to be putting tape on my $8000 camera to fix what should have been aligned correctly to begin with (and losing my 100% viewfinder in the process. . .)



That, and their refusal to use what is now commodity sensor-based antishake which breathes a huge amount of usefulness into fast primes in low light. . . Admittedly Nikon is playing the same game here with their customers. . .

In any event, you should be thankful that Sony is about to introduce some competition into the high-resolution full-frame dSLR market. It will make your Canon bodies a lot cheaper in the future. . .

I've seen evidence that the 1Ds3 alignment problem between the viewfinder and the image may exist. However, I've done a series of hand-held tests with my 1Ds3 this evening, photogaphing rectanglar furniture and a painting on the wall, and I have not detected an alignment problem even with this crude kind of testing. It may be argued that the testing is too crude to replicate the problem, but to cover for that I made a number of exposures, and where I succeeded to get the image aligned in the viewfinder it wasalso aligned on my display.

As for the auto-focus issue - again I cannot replicate this problem on my 1Ds3. In fact it focuses in almost total darkness, and it focused perfectly on very bright snow scenes here in Toronto last week.

Am I lucky to have gotten a good one, or are there a few unlucky people with sub-par ones? We'll know with more evidence.

As for Canon not being conscious of the need to market a top quality product for top dollar - I simply don't believe it. They may be insular and aloof, but many millions of cameras later they aren't dummies either and they MUST know what is going on around them - including all the rumours - because if you guys know these rumours you can bet your bottom dollar they knew them well before any of us, yet their prices are their prices and once they price the current model where they did, it's unlikely they will reduce it - that would be for the next model which may well be a good year or two away; meanwhile they need the current model to survive commercially. Either they simply couldn't price the 1Ds3 lower than they did, or they could but they aren't concerned about prices from Nikon and Sony. Time will tell.

As for Sony and Nikon, doesn't Sony manufacture Nikon's sensors? None of these folks live in a vaccuum. Competition is badly needed in this segment of the industry, but it won't be totally arms' length, and it won't levelize prices for differentiated products with many people holding expensive and dedicated legacy accessories. A 20+ MP camera isn't quite a commodity yet like a pound of copper.

Lightstand,

Are you saying you never had a problem with the alignment between the film guides and the viewfinder of a view camera? How could you? They don't have viewfinders. They're called "view cameras" because you "view" the actual projected image from the lens, in exactly the same place the film will be. Maybe you mean the film holder guides in the cameras have always been parallel with the mounting surfaces of the cameras?

The OP is talking about the viewfinder image not coinciding with the sensor image. This should be extremely easy to test. Put the camera on a tripod in front of something with parallel horizontal lines. Line up one of those lines with one of the edges of the viewfinder frame. Take a picture, or better yet, activate live view, and see if lines in the scene are parallel to the image borders.

My old Canon S50 point and shoot is out of alignment a lot, but I've detailed my simple solution earlier. With my DSLRs, I don't see how the viewfinders are big enough to align an image to within one degree anyway. I'm really interested in seeing how big this discrepancy is with 1DSmkIIIs.

I just want to report that both my cameras serials something with 605xxx are absolutly fine. I checked both carefully on my computer screen and couldn't see any difference between live view and the viewfinder. That is great and how it should be.

I do understand your line of reasoning, but the test seemed to indicate that the whole assembly (viewfinder, screen) may be out of sync. Line up in the viewfinder with the grid lines (or the edge of the frame).... and the resulting image is rotated. Even if your fix did work, it doesn't really help as, like so many photographers, I compose in camera to a great extent and the edge of the frame is an important part of that process.

But any attempted string and sticky tape solutions simply don't get away from the fact that it in such an expensive piece of kit things should work perfectly - out of the box!

It is not rocket science (or attributed to the the cutting edge nature of the new technology), it is simply nailing the bits together correctly!!

Maybe one misaligned camera may come off the production line in Japan on a Friday afternoon following a long lunch of sushi and too much sarke - but I've personally seen four with serials spreading over hundreds of units.

I've waited a long time for this camera to arrive. I actually want it to work!

OK, just did some tests in my house with both 1Ds2 & 1Ds3, using a wall with a couple of paintings & doorway, and I do get some noticeable misalignment I would normally need to correct, not noticeable in my 1Ds2 which actually shows a bit less of the actual frame than my 1Ds3, though well aligned.

My 1Ds3 S/N is 615xxx for reference, and no I'm not switching back to my 1Ds2 - I much prefer using the somewhat misaligned 1Ds3. In my case I can deal with it until Canon hopefully comes up with a fix.

I just bought a new Bently, but like all the recent models it has a duff starter motor, but as I always park it facing downhill it renders the issue moot. Though I gathered if you got a starting handle and placed it under grill and yanked it a few times it would normally start.
God, you get some daft advice online sometimes.


Yesterday in my local dealer they just had a 1DsIII returned and the other 3 in stock were also useless. So looks like Nikon are going to be selling even more high end cameras than they could possibly have imagined as both the recent 1d + 1ds arrived with serious quality control issues. And it's not as if Canon [UK] are any good at sorting normal repair problems out either. For pros or amateurs.

I apologize for the vague reference it was aimed not at the OP but the idea that the misalignment is because the simple "over-priced" bubble levels could never level a camera. A sentiment I do agree with to some extent however I do know with my 1dsm2 a misalignment between viewfinder & sensor is present.

Just got the grid screen the other day and mine's out 0.75º.

A hassle but hardly end of the world stuff. I'll get onto CPS tomorrow.

Yes, like "the hood ornament on my Mercedes appears to be misaligned clockwise by 0.75 degrees, so I'm going to return the car and walk 20 miles to work..."

For a 0.75-degree misalignment, re-seating the grid screen in its' socket will probably fix the "problem". And I'm still skeptical than anyone can realistically "eyeball" level to that degree of precision, even with a grid screen.

Hi,

A week after I got my Ds3 I had a few weeks in Maui to do some shooting. I started to have this sensation that my viewfinder and bubble level did not agree. (It was a bit disorientating!)

The live view looks right so I am assuming that it is something with the viewfinder.

Now that I know that I am not the only one I will just sit back and wait for a fix to be announced. My focus tracking has been A1 in the hot and bright Hawaii sun.

This is one great camera for image quality and a significant refinement over the Ds2. I hope postings that trash this camera do not put people off of experiencing the excellent images one can get. It is not just less noise with a larger image file, but the quality of the noise that is there is "nicer" and more "film like" This really shows in B&W conversions and allows more extreme channel mixes without excessive and irritating noise and artifacts.

My "sub par" 16-35f2.8L is the best wide I have ever used. Maybe there is something even better out there but this lens is no dog. If you want to see bad, look at the corners of the famous 40mmCF Zeiss lens for the Blad, (one of my seven blad lenses). The Zeiss name is no mythical guarantee of optical nirvana as some posts would indicate. When people are throwing feces around, more seems to stick to the outside of Canon lens barrels than to the teflon coated barrels that the Zeiss lenses must have.

In terms of some one like Sony placing a full frame sensor based image stabilization system in there bodies, I am not sure of the physics. Would not the coverage of "full frame" lenses not be enough to cover the shifting of the full sensor?

Jonathan,

I don't believe it is the grid focussing screen that is mis-aligned. In one case I know of the camera is equipped with the factory screen, yet the image is rotated .75 degrees. The test was repeated with a 5D - same setup, same lens etc - and no rotation was observed.

The importance of an accurate viewfinder is really an individual decision for each photographer. For example, I shoot with tilt-shift lenses and use the Angle Finder C to assist me in aligning the images in the viewfinder. So, for me, the accuracy of the viewfinder is crucial - I would be unable to fully utilize my 1Ds3 if there was a .75 degree viewfinder error built into it. I've returned items for exchange to my dealer in the past for various flaws discovered shortly after purchase, and I don't see why this would be any different. The decision is the purchasers - I don't think it unreasonable.

For the record, my own 1Ds3 does not appear to have this problem. I currently know of an additional 1Ds3 owner without the issue as well.


Stephen

Again, my experience in the surveying world leads me to believe Jonathan is correct about this. Are any of you who are experiencing this problem absolutely certain that it was never present at all in any of your other cameras?

I never obtained (or needed) that kind of framing precision in the film days, but I suppose there would have been no point when one could simply rotate the print under the enlarger without even thinking about it. But to fractions of a degree?

If your work requires such precision, are you sure that taking a few seconds to rotate the image in Photoshop is really that disastrous? Aren't there a thousand other things you have to do to produce the finished results, making this just one tiny part of the puzzle? I'm not criticizing anyone here, just trying to fully understand the issue.

Some of you may be misunderstanding what Jonathan is suggesting with his tape solution. He isn't suggesting you tape over part of the viewfinder or the focus screen. What he's saying is that you may be able to rotate the focus screen so that it is then exactly parallel to the sensor. You might force that rotation by adding a tiny piece of tape to the outside of the screen frame on one corner. Essentially shimming it a tiny bit so that it turns inside its socket.

I did a quick back of the enevelope calculation, and I'm thinking that the required shim thickness to correct the .75 degree error would be .0047 milimeters. I don't have a 1DsMIII, so I can't know if my calculation was realistic, or even how the assorted viewfinder parts work together, so I could be off a good bit.

Assuming I'm even within an order of magnitude on that, I think it's pretty clear that we're talking about extremely fine measurements. I'm wondering if that kind of tolerance on a replaceable part like a foucs screen is even a realistic manufacturing possibility.

Misirlou,

It doesn't appear to be the screen.

In the one instance where I've seen evidence of the misalignment, the camera had the standard factory screen installed (not the grid screen). A test chart was photographed first by aligning using LiveView. The test chart was then photographed by aligning with the viewfinder frame and the resulting image was .75 degrees rotated. The test was followed by shooting the target using a 5D, without a grid screen, aligning with the viewfinder frame and there was no rotation of note. I don't see how shimming the screen would affect the end result in this case.

A grid screen is installed in my 1Ds3 - which does not appear to have the issue.

S.

I will say this, I love my 1Ds3, nothing will make me go back to my 1Ds2 except an emergency or where I really need two cameras at the same time, but I don't need a level to tell me my viewfinder is off by probably 3-4 degrees, noticeable if you try shooting straight doorways, wall sides, any obviously perpendicular or horizontal shape. If you turn on Live View it's immediately obvious if you aligned a vertical or horizontal line in the viewfinder.

I just went through a test with both 1Ds2 & 1Ds3 and the discrepancy is visible in the Mk III without any measurement, just the eyeball, and any working eyeball at that, no experience necessary. Still went out and had a great time (and some nice shots) doing marshes and trees and such.

I live about 20 minutes from Canon's headquarters in Lake Success, NY. If anybody knows somebody there I'd be happy to show them my 1Ds3 and they'd see it immediately, or else they're not driving legally .

A further test shows it to be the v/f, rather than the screen.

Test One: Centre horizontal line of the grid screen placed exactly along the matt / print border. The "slope" is clearly seen, even of the camera's display.

http://www.philaphoto.com/images/test1.jpg

Test Two: Just tilted the camera up so the same matt / print border was just above the bottom edge of the v/f. I didn't worry about getting square on for this one.

http://www.philaphoto.com/images/test2.jpg

Hardly super scientific but shows the problem.

Michael,

The 5 degree error reported, after further detailed testing, turned out to be a 1.6 error, however still unacceptable.

Nick Walker

Shimming the screen is only effective if you have a grid focus screen installed, and you align your subject to the grid instead of the viewfinder edges. Even if the mask on the viewfinder prism was misaligned slightly, it's quite possible that the focus screen mount is not misaligned, and if so, installing a grid screen would provide one with an accurate level reference without shimming the screen at all. It's a cheap and easy fix to try, and much less hassle than returning multiple cameras.

Jonathan, first, as I mentioned above, I don't appear to have this problem with my 1Ds3. Having said that, clearly there is an issue, but we don't know how widespread. I would also suggest there could be a number of explanations for it. Only Canon engineers would know for sure. So while your fix may work based on your hypothesis of the cause - and it is helpful that you are suggesting people with the problem should try these things to see whether they work - this could well be a quality control issue at the manufacturing plant in Japan. Canon clearly knows how to manufacture cameras to near-perfection, but it wouldn't be the first time that silly things slip through their procedures and end-up on the market. I really think the way forward on this issue is for people who have the problem to send Canon carefully prepared evidence allowing them to investigate the issue. They aren't always the most responsive corporate entity around unfortunately, but my sense is that on something as basic as this on a flagship product entering an intensively competitive environment they will respond, as they did on the 1D3 focusing issue. It may take some pressure, but in the final analysis the only correct solution is for Canon itself to repair or replace any defective units, and not leave it to user-concocted workarounds. That will not enhance their reputation as a world-class camera manufacturer.

I'm not saying that Canon should ignore this issue. If someone wants this issue corrected, Canon should do so under warranty. I just think that some of the reaction to it (it makes the camera completely useless, one should sell all their Canon gear and buy Nikon or Sony, etc) is hyperbolic BS. It's hardly in the same class of problem as the shutter failing, or the camera locking up unexpectedly, etc. Let's keep a bit of perspective here. I'd gladly trade a slightly misaligned viewfinder for all of the other improvements a 1Ds-Mark III would offer compared to my 1Ds-Mark I.

I agree.

Jonathan, you mentioned earlier in this thread that you are in the army.

Would you accept going into battle with a weapon that doesn't shoot where you've aimed it? .... I doubt it.

As a professional photographer, when I point a £6000 digital camera at a subject I expect to capture that subject in all it's glory - not slightly sloping with a bit missing off the corner - period.

As I mentioned before, I've personally tested and rejected FOUR of these cameras - and many more people are experiencing the same issue. I'm now beginning to wonder if they all suffer from the same problem to a lesser or greater extent - sloppy construction.

Wow. That is pretty extreme. About as misaligned as my old S50 point and shoot. I suspect there is something about these cameras that allows a critical component in the viewfinder optics to rotate when it shouldn't. I'd be surprised if Canon didn't offer some sort of fix pretty quickly. If I had one of these cameras, I'd use live view and a tripod for things that need critical framing, at least until Canon comes up with a solution.

Carl, were the four cameras you rejected from the same batch - judging from the serial numbers?

I have just done a further test on mine, and I've come to realise this is not easy to do. Even (or perhaps especially) on a tripod, it's hard to get a perfect rectangular alignment of a perfectly rectangular subject. I had better luck last night hand-held. Anyhow, FWIW, last night's image showed no perceptible issue, hand-held aligning the image (of a painting on the wall) in the viewfinder. This morning's image on the tripod also aligned through the viewfinder showed a mis-alignment of 0.18 degrees CCW on two dimensions top and left vertical) and 0.41 degrees CCW on the two other dimensions (bottom and right vertical) - but measuring this also means placing the ruler very exactly where it should be left and right or top and bottom. I did this at 100% on edges that have decent contrast.

I guess my position on this is that if Canon issues an advisory, I'd send them my camera for precise testing under laboratory conditions and adjustment as indicated. Meanwhile, I shall continue using it happily. I find that most of my shots involving straight horizontals and verticals often need some kind of perspective and de-skewing adjustment anyhow, so this is not a deal-breaker even if there were a slight mis-alignment of a component in the camera.