 |
 | Home | Contact Us | FAQ | Search & Site Map | Link to Us |
 | Sign In | Join | Other 45 Sites in Network |
| PhotoKB HomeDiscussion Groups Digital Photography Digital PhotoDSLR CamerasZLR CamerasPoint & Shoot CamerasFilm Photography 35 mmLarge FormatMedium formatDarkroomFilm and LabsOther EquipmentPhoto Technique Nature PhotographyPeople PhotographyTechnique GeneralGeneral Photo Topics General TopicsAustralian PhotographyUK PhotographyDirectoryPhoto Clubs |
Photo Forum / Digital Photography / DSLR Cameras / September 2006Focus calibration
I believe my 20D's focus is off a bit. With the same lenses, I get sharper images on my old Digital Rebel 300D then I do with the 20D under similar circumstances. I have seen several testing methods and will probably go with a yard/meter stick placed on the ground (zero pointing at my tripod and max pointing directly away from the tripod), camera mounted on a tripod, at around 45 degree angle, about 8 to 10 feet away with the focus point (single, center) on the mid-point of the stick. Using the timer and mirror lock-up and shooting a variety of apertures at 100iso should yield some demonstrative results. I plan to do this with both cameras (300D & 20D) to validate my perceptions as well. The question is, once I do all of this and the tests bear out my perception, how do I go about correcting the calibration? I naturally assume Canon or a Canon authorized repair facility will need the camera (no longer in warranty) but I don't know the following: How long will the repair shop have it? Expectations of repair effectiveness? Cost? Contact Canon directly or take it to a reputable camera store? (Washington, DC area, Maryland side...)? Do they need the lenses I use as well? Should the adjustment be good for future lenses (I am assuming yes...)? Any thing I am missing? If anyone has any experience with this sort of calibration, I would appreciate your input. Thanks > I believe my 20D's focus is off a bit. With the same lenses, I get > sharper images on my old Digital Rebel 300D then I do with the 20D under [quoted text clipped - 3 lines] > mounted on a tripod, at around 45 degree angle, about 8 to 10 feet away > with the focus point (single, center) on the mid-point of the stick. <snip> Be aware that the central focus point sensing area is considerably larger than the little box in the viewfinder would indicate, and presenting the focusing sensor with a 45-degree angled subject will not guarantee good focus. For accurate focus checks, the area focused on should be larger than the sensor, and square to the camera, and of reasonably high contrast. The manual for the camera will most probably have an illustration showing the actual area covered by the focus sensor. Any conclusions from your proposed method will be invalid. Colin D.  SignaturePosted via a free Usenet account from http://www.teranews.com >I believe my 20D's focus is off a bit. With the same lenses, I get >sharper images on my old Digital Rebel 300D then I do with the 20D under [quoted text clipped - 22 lines] >If anyone has any experience with this sort of calibration, I would >appreciate your input. Thanks I test all my lenses and camera bodies using a picket fence or a set of iron railings. I put a glove on the fence, position myself close to the fence and focus on the glove. It is not worth doing this at any aperture other than wide open. It is definitely worth doing this at several different focusing distances. I have had my Canon 5D's focus recalibrated twice; once when new, when it was clearly slightly off straight from the factory, and once when my camera bag took a tumble. I took it to a Canon authorised repairer in London, England. Both times took less than 24 hours. Both times I took along my 50mm f/1.8 and 24-105mm f/4 lenses. > I believe my 20D's focus is off a bit. With the same lenses, I get > sharper images on my old Digital Rebel 300D then I do with the 20D under [quoted text clipped - 12 lines] > > How long will the repair shop have it? probably very little time. 1 or 2 days in the shop + shipping there and back. But call them. > Expectations of repair effectiveness? unknown. > Cost? Probably outrageous ($300?) to check it over. > Contact Canon directly or take it to a reputable camera store? > (Washington, DC area, Maryland side...)? > Do they need the lenses I use as well? unknown, but again, ask them. > Should the adjustment be good for future lenses (I am assuming yes...)? > Any thing I am missing? Your comparison cameras may be set to different amounts of default sharpening, in which case the Rebel may look sharper. You might try to compare manual focus to autofocus. Maybe first try a lens like 50mm 1.8 (an excellent and quite low cost lens) also much brighter than the typical zooms. Unfortunately the screen on the 20D is hard to use for judging manual focus. Also check infinity focus. If you compare photos from one camera to another, make sure to use a tripod for both. > If anyone has any experience with this sort of calibration, I would > appreciate your input. Thanks Since you will not be able to change the focus calibration yourself, this will be an unhappy situation if you determine that focus is a problem. > I believe my 20D's focus is off a bit. With the same lenses, I get > sharper images on my old Digital Rebel 300D then I do with the 20D under [quoted text clipped - 22 lines] > If anyone has any experience with this sort of calibration, I would > appreciate your input. Thanks I just got my 350D back for the second time. The right side of the frame was softer focus than the left. They were not capable of understanding English because they never did solve the problem. When I got the paperwork back with the camera they had it exactly opposite of what I said the problem was. They said left side was softer. Didn't matter because they didn't fix the original problem anyway. I sent pictures and explained it in great detail. I don't think they read what I wrote and couldn't have looked at the pictures. They did however find an overall focus problem (after sending it back a second time) and now the camera focuses very well but the right side is still slightly softer. It was under warranty at the time. Bottom line is don't count on them solving a problem the first time. They seem to know how to do basic repairs but can't comprehend there could be more serious problems. In you case a simple calibration might be all you need, but it took two times for them to find the overall focus problem. > I just got my 350D back for the second time. The right side of the frame was softer focus than the left. They were not capable of > understanding English because they never did solve the problem. When I got the paperwork back with the camera they had it exactly [quoted text clipped - 5 lines] > comprehend there could be more serious problems. > In you case a simple calibration might be all you need, but it took two times for them to find the overall focus problem. This type of misalignment is FAR more common in lenses (especially zooms) than it is in bodies. Do you have more than one fairly short FL lens, and is it also softer on the same side of the frame? If so, it is likely the body... -- David Ruether ruether@verizon.net rpn1@cornell.edu http://www.ferrario.com/ruether >>I just got my 350D back for the second time. The right side of the frame was softer focus than the left. They were not capable of >>understanding English because they never did solve the problem. When I got the paperwork back with the camera they had it exactly [quoted text clipped - 9 lines] > than it is in bodies. Do you have more than one fairly short FL lens, and is > it also softer on the same side of the frame? If so, it is likely the body... Not so. The misalignment of sensors, and AF modules, is almost universal in bodies. Lenses are comparatively easily made to be accurate. Bodies have firmware registers programmed to indicate their degrees of error in collimation - lenses don't! Fixation, a London company, specialises in setting up DSLRs (mainly Canona and Nikon) to an accurate standard of adjustment of sensor plane and AF module positions, to reprogram for close to zero offsets in the firmware tables which adjust lens focus (imperceptibly) to compensate for the errors found in DSLR bodies. I've spoken at length with their MD and studied two service manuals and confirmed the facts - the problem lies mainly with getting any sensor assembly to sit in the focal plant, perfectly, and secondly with getting a prefabricated AF module into exactly the right place. David [] > Not so. The misalignment of sensors, and AF modules, is almost > universal in bodies. Lenses are comparatively easily made to be [quoted text clipped - 13 lines] > > David Isn't this sort of once-off alignment something you expect to be done at the factory? Perhaps it drifts after setting as the various components bed down? David > [] > [quoted text clipped - 19 lines] > the factory? Perhaps it drifts after setting as the various components > bed down? Each camera has its program done at the factory on a testbench. But, as one poster has commented, it only takes a minor knock to seat the sensor assembly into a different position. The problem is that the sensor is not pressed up against a film gate, it 'floats' on a circuit board mounted on three adjustable posts which between them control focus plane, pitch and yaw. There is a further adjustment for rotation, as the sensor can also be misaligned with the horizon as seen through the viewfinder. Minolta, for example, permitted up to 0.5 degrees of rotational error - try that in Photoshop and it's a big error! When a camera is repaired, the entire process must be repeated on a service test bench, using the dedicated software and special tools. Some Minolta calibration lenses (they look like 50mm f1.7s and 75-300mm zooms) have appeared on eBay recently, since KM shut down. Looks as if their attempts to ensure the skips of spares were all crushed and destroyed failed, and these items have been 'rescued' by scavengers, who don't know what they are. It is possible that the test targets, light boxes, benches and PC software, connection cables etc may also be out there at car boot sales. A pity, because without this stuff, it is impossible to recalibrate the metering and AF of pre-digital bodies; and the focus accuracy and sensor rotation etc of digital bodies. People who have done 'their own back focus adjustment' by peeling off baseplate assemblies and accessing adjustment screws, or who have adjusted secondary focus mirrors etc (common with Nikon) can't of course alter the firmware programmed offset tables which tell the camera's CPU the extent of errors at a range of settings. This is where Fixation in London scores - they can take pro Nikon and Canon bodies, and set up the sensor pitch/yaw/focus/rotation, mirror resting alignment, focusing screen shims, secondary mirror stop, AF module pitch/yaw/focus; then reprogram the camera so its table of offsets is as close to zero as possible, which speeds up response by bringing these adjustments within the range of error which the camera treats as zero. There is then no tiny 'flick' of the lens focus between pressing the shutter button, and the shot being taken, which happens when the calibration table indicates that the focus must be adjusted. And the overall accuracy of shots, sharpness etc is apparently improved to the extent that all of the major newspaper photo depts, news agencies, and pro shooters who are aware of the service get it done. Apparently it's a good idea to have it checked and adjusted regularly even for the most expensive pro DSLRs. Sensors can't be relied on to stay in place. David [] > Apparently it's a good idea to have it checked and adjusted regularly > even for the most expensive pro DSLRs. Sensors can't be relied on to > stay in place. > > David Thanks for the detailed explanation, David. I'm sure that the alignment issues will apply to cameras other than DSLRs, although whether paying to have them adjusted regularly (even if it were possible) is unlikely to be justified. I would therefore take the view that if a new camera isn't correct out of the box, return it immediately. Cheers, David > People who have done 'their own back focus adjustment' by peeling off > baseplate assemblies and accessing adjustment screws, or who have [quoted text clipped - 11 lines] > pressing the shutter button, and the shot being taken, which happens > when the calibration table indicates that the focus must be adjusted. David, To make sure I understand you, are you saying that the focal planes of the groundglass, AF sensors and imaging sensor are at different distances (ie that if the image is focused on the groundglass it might be a bit off for the sensor)? This I can easily believe, and also that this difference is stored somewhere to be adjusted for while autofocusing. But are you saying that the focus distance is shifted after pressing the shutter release? I find this rather hard to believe, for 2 reasons: 1) In a 8 fps second camera (or even 5 fps), this would not be trivial to do, I think; 2) The focus confirmation indicator would give wrong results for manual focus lenses (unless in that case it gives confirmation only when the image would be focused on the sensor, rather than on the groundglass, and someone would presumably have noticed this by now; although maybe the difference is too small to see through the viewfinder). Anyway, I'd find it more plausible that the difference in length of the two lightpaths (to sensor and to AF sensors) is stored, and this is used to focus such that the image is formed on the sensor. But I don't actually know, of course, so if you have reliable information, I won't insist. Or is that what you were saying? Although, come to think of it, if a small correction is indeed made after the shutter release is pressed, and this is done "ballistically", ie if the lens itself somehow can tell the camera by how much to turn the screw for a given change in focus, this might explain how systematic AF errors can depend on the lens and not the body (something I never understood). But how could this explain anything for lenses with integral motors? Maybe in this case the camera tells the lens "must adjust focus by x amount" and the lens then looks up in a table exactly what this corresponds to (in terms of voltages to motors etc) and does it. Then if this table is off, everything is explained. Well, except the business with MF lenses. Which is also explained if the AF system then just tells you that it's in focus when it really is; this should be easy to check by using AF to focus with a fast lens, then switching to MF and seeing if the in-focus indicator remains on. Should do that. > To make sure I understand you, are you saying that the focal planes of > the groundglass, AF sensors and imaging sensor are at different [quoted text clipped - 12 lines] > by now; although maybe the difference is too small to see through the > viewfinder). Maybe 'during' pressing the shutter release is more accurate. Yes, the AF system gets the focus perfect and the lens either backtracks or continues as the mirror rises, to a position which is not exactly that of the AF focused point. This is why the press guys get their DSLRs blueprinted. It does slow down motordrive, even at 3 fps, let alone 8 fps, because the lens will refocus again after each shot. Since the Nikon D80 and my Sony A100, and both my KM 7D and 5D, all refocus the lens on a perfectly static subject if AF is left active when shooting sequences - just a tiny amount, it differs from camera to camera - I assume all four are using a programmed offset for the focus. > Anyway, I'd find it more plausible that the difference in length of the > two lightpaths (to sensor and to AF sensors) is stored, and this is [quoted text clipped - 3 lines] > > Or is that what you were saying? You are not going to like the facts. The difference in real distances is programmed in, but not all lenses can be adjusted to match. This is why for example Nikon DX (G) CPU lenses, or Minolta (D), or the latest Canon EF-S lenses, may be less prone to backfocus, front focus etc than for example a nice vintage early 1990s 50mm f1.4 or whatever. The latest lenses have more accurate distance encoding and can be adjusted better. As for the focusing screen, most makers adjust its accuracy in relationship to the primary mirror, which itself is adjustable in stop position, using between one and three metal shims. The bad news is that the AF markings - the things which light up, and which you tend to focus your eyepiece on using the dioptre control - are often on the bottom of the prism/mirror unit, and the focus screen then sits below these, and is separated from this plane by its shims. So adjusting the dioptre for dead sharp view of the focus markings may not mean you are focusing on the screen itself, and with modern semi-clear screens, that can lead to poor manual focusing accuracy. > Although, come to think of it, if a small correction is indeed made > after the shutter release is pressed, and this is done "ballistically", [quoted text clipped - 11 lines] > switching to MF and seeing if the in-focus indicator remains on. Should > do that. Some people complain that the camera locks focus and it doesn't look sharp in the finder - that may be so. It does not follow that the result will be unsharp. I spend ages fine-tuning focus on the focusing screen, over-riding what I thought was bad AF, before realising that half the time I was over-riding a correction which IMPROVED the focus, and what I saw on the screen was wrong. However... You can't focus manually on a normal DSLR at all. AF focus indicators can be used, but current focusing screens are not suitable for manual focusing. Konica Minolta sold a retrofit type M and type ML (grid) true groundglass for the 7D, which enabled accurate manual focusing and gave a good indication of depth of field. Maybe other makers offer the same thing. The metering always has to be recalibrated, as these less transparent screens lose about 1.5 stops of light. All the makers fit next to useless focusing screens - no matter how many howls of complaint saying 'I can focus fine with mine!' arise. These screens are coping with a 70 per cent light loss caused by the secondary mirror system diverting maximum light levels to the AF sensors, plus a small prism or even worse a mirror prism, a low aperture ocular, low magnification factor (the Canon EOS 300D for example shows a view which is really 0.46X in 35mm system terms); then add to this modern zooms with maximum apertures as low as f6.3 at many focal lengths. If you put all this in with a real, proper, interceptive groundglass it would be so dark no-one would ever buy the camera. If you have a modern 1.5/1.6X factor DSLR and a 50mm f1.4 lens, here is simple test to demonstrate the problem. Set up a still life at about 18 inches distance, with a well lit background maybe five feet away, and plenty of foreground and other detail - lots of depth. Light it well, brightly. Focus your DSLR with 50mm fast lens. Examine the apparent depth of field through the finder, if you have d-o-f preview, do this at several settings - f2.8, f5.6, f8, f11, wide open, etc. Shoot some pictures using these settings. Now compare the depth of field on your shots, by reviewing on the camera screen, with what you see through the finder at the same apertures using d-o-f preview. Anyone using a typical f4-5.6 lens won't be as aware of this situation. Below f6.3, say f8 or f11, the d-o-f matches up well. The big problem arises with faster apertures like f2.8 or f2. DSLR viewfinders can be extremely deceptive and lead to confusion, disappointment and misjudgements of depth of field with fast lenses. David photoclubalpha.com > This is where Fixation in London scores - they can take pro Nikon and > Canon bodies, and set up the sensor pitch/yaw/focus/rotation, mirror [quoted text clipped - 14 lines] > > David To bad Canon is not capable of the same. >> This is where Fixation in London scores - they can take pro Nikon and >> Canon bodies, and set up the sensor pitch/yaw/focus/rotation, mirror [quoted text clipped - 17 lines] > > To bad Canon is not capable of the same. I met the Fixation MD on a Nikon press trip! He was along because of the valuable work his company does tweaking pro Nikons. Canon actually will pay for a camera to be adjusted, and Fixation do this. You have to show that it's not good enough. I do not have Canon's tolerances, I only have Konica Minolta ones to hand, because I bought CD-ROM copies of the service manuals for two cameras from Peter Ganzel. I would bet they are about the same. I can sum it up as 'if it's within 20 per cent of the target value it does not need repair unless the customer insists' - then it gets adjusted until it's within 10 per cent. You and me, we want 5 per cent. And somewhere out there, a few owners will have DSLRs which just happen to be within 1 per cent, by chance. But the regular punter won't get a free repair unless the target shot shows resolution is down by 30 per cent, or exposure is more than 1/3rd stop out, or the sensor is out of horizon level by more than 1 degree... etc. David I can > sum it up as 'if it's within 20 per cent of the target value it does not > need repair unless the customer insists' - then it gets adjusted until [quoted text clipped - 8 lines] > > David What do you mean by "target value"? On my camera the right half of the frame is softer focus than the left getting worse the further right you look. It is slight but still a problem at times. It is especially a problem when shooting a person vertically and their head is in the top third of the frame. (Which would actually be the right side of the frame if looking at it horizontally, if that makes sense)If I turn the camera around they are very sharp. I really noticed this when I found I was doing more sharpening than usual on pictures taken vertically. It's probably within their spec but I don't feel I should have to turn the camera to an awkward position to take sharp pictures. If it can be sharp on one side it should be just as sharp on the other. > I can > [quoted text clipped - 24 lines] > the camera to an awkward position to take sharp pictures. If it can be > sharp on one side it should be just as sharp on the other. What I mean is that the camera is placed on a test bench and a special certified lens is fitted, prefocused on a target. If the image produced by the sensor is within X degrees of straight, and X deviation from correct sharpness of the target at five measured points, it passes. If it's outside these QC figures, they adjust the camera. There is a tolerance, within which a camera leaving the factory is deemed OK, and one returned for check or repair is reported as being 'within specifications'. You are not alone, most sensors are not perfectly plane parallel to the lens - they are after all a circuit board floating on a bit of electronic mounting attached to a back plate in the camera by adjustable posts. I had a Kodak DCS Pro/n where the moment you mounted any sort of real wide-angle, like the 12-24mm Sigma, one side was out of focus. Answer - it was never made to be accurate enough for a 12mm lens. David >What I mean is that the camera is placed on a test bench and a special >certified lens is fitted, prefocused on a target. If the image produced [quoted text clipped - 12 lines] >real wide-angle, like the 12-24mm Sigma, one side was out of focus. >Answer - it was never made to be accurate enough for a 12mm lens. David, Thank you for your fascinating and hugely informative contribution to this thread. Tony |
Reply to this Message |
 Sign In
 Join
 My Latest Posts My Monitored Threads My Blog My Photo Gallery My Profile My Homepage Start New Thread Enable EMail Alerts Rate this Thread
|
©2009 Advenet LLC Privacy Policy - Terms of Use
This website includes both content owned or controlled by Advenet as well as content owned or controlled by third parties.