How does it compare to pushing color negative film to 3200?

Craig,

Have you spoke to Canon about this? I haven't talked to them yet, but the more testing I do, the more obvious it is that there is a problem.

What I still can't figure out is if everyone has this problem and just don't notice it, or if this is a problem with a few cameras and something Canon can fix.

Like I mentioned before, I never saw this probel with my rebel, even though the noise was fairly pronounced sometimes at 1600.

RDKirk wrote:
How does it compare to pushing color negative film to 3200?

What's your point, exactly? Color negatives don't suffer banding issues and do not warrant any mention in this discussion. Color negatives may indeed have more "grain", but that's an entirely different discussion.

What we have in some cases with banding is a sensor that will show different bandings characteristics shot to shot with identical camera settings. You'll be hard pressed to convince me it isn't an issue Canon should try to address.

Sam,

I sent you two full sized pictures to check out. The second one I pushed about half a stop to show the banding a little better. Let me know what you think.

When you see banding are the bands always the same width? I wonder if this is an integer number of pixels whether a relatively simple chunk of code (or possibly even a PS action) could be written to address this. Artifacts with well defined periodicity are sometimes the easiest to remove.

Granted Canon ought to fix this, but if we can do it ourselves then I wouldn't get too bent out of shape over it.

I just got my 20D in the mail, pumped it up to 3200 iso, exposed at 30 seconds and 20f with a 2nd curtian flash (on camera), put it in the bathroom, shut the door, and had banding all over the place! Im using a 50mm 1.4 on manual. On top of that, the grip for the 20D is a bad design... do any of you feel the same?

ok, so I did a whole bunch of tests. Here is my conclusion. The banding occers only when you use the on-board flash. The lower the iso the less the banding. An external flash produces no banding, and using no flash at all also produces no banding. The is fine for those who will only use external flases, but when I travel with friends, I want to be able to use the on-board flash! I have not tried to do this with the noise reduction on. Mabe tomarrow, canon will be getting a call... let me know if any of you find the same results.

For what its worth,
I got the exact same problem with my 10D shooting indoors at 3200ASA (at a wedding) Canon replaced the sensor under warranty. While it was in for repair the shop gave me a loan of both a D60 and then a 10D and both cameras exhibited this effect under these types of conditions.
I’m sure someone in Canon could explain it but I suspect it is a function of how the senor and software intrepid the incoming info under these circumstances.

I've never seen banding at any other ASA setting on any canon.

There's someone on dpreview who is having this exact problem with his shiny new 1Ds Mark II:
http://forums.dpreview.com/forums/read.asp?forum=1032&message=11095080

One respondent has suggested a faulty oscillator, it certainly seems like there is some sort of electrical fault causing the problems.

Steve

This would invalidate Chris Moon's test result. I doubt that anybody would use flash for astronomical shots . By the way, the DPreview photograph looks horribly overexposed (or underexposed and overcorrected).

I meant that the type of banding seen in the image is exactly the same, not necessarily the precise cause.

The common thread between all the examples appears to be some sort of electrical factor. Onboard flash has done it in some cases, in others the autofocus motor of a specific lens, in the case of the person on dpreview it seems to be afflicting every shot.

So, the most likely explanation, to me, seems to be that there is an electrical fault that can be present with varying degrees of severity. Some cameras never have any banding, some have it all the time, some are in between and need an additional contributing electrical factor to cause it. High ISO is always going to be more revealing of such problems.

Steve

I agree. So it seems it is time now for the electronics experts to step in!

jdee's problem definitely seems similar to mine, but much much more pronounced. The frequency and width of the bands also varies across the image, which is quite odd. The shot settings are as follows - ISO 200, 30 seconds, f/7.1, Second-Curtain Internal Flash:

Yep, long exposures are another thing that will reveal any electrical problems, as well as high ISO.

My advice, for what it's worth, is that anyone who has this problem should assume, until it's proven otherwise, that the fault only affects a small number of individual 20D copies, and hence they should ask for a replacement for their defective camera. If you find the same problem on the second and third cameras you try, a rethink will be needed.

Steve

Send it back, needs to be repaired by Canon. Sorry Dave

stevei wrote:
My advice, for what it's worth, is that anyone who has this problem should assume, until it's proven otherwise, that the fault only affects a small number of individual 20D copies, and hence they should ask for a replacement for their defective camera. If you find the same problem on the second and third cameras you try, a rethink will be needed.

Steve, my advice to you is that before recommending someone enter the often grueling Canon Warranty Loop of Hell you take the extra 5 minutes to see if perhaps your camera is exhibiting the same problem. I did, and, well... it does.





And yes, I realize I need to clean my sensor.

Edited by Sam Bennett on Nov 17, 2004 at 11:09 AM GMT

I advised replacement, not repair

I agree, Canon warranty is a nightmare, at least it is in the UK. I'd be asking my retailer for a replacement camera, they're not in particularly short supply any more.

I don't have a 20D, though I'm intrigued enough that I'll try it with my 300D this evening. I was assuming that if it were a common problem, then a lot of other people would have seen it too. On further reflection, though, I suppose there won't be that many people using the internal flash on long exposures, so it might be a common problem that the vast majority of people never experience.

I was also working on the basis that getting a replacement camera from the retailer is very little hassle, but I suppose if you've bought it over the internet then swapping it might not be so easy.

Steve

Are you shooting in RAW or not?

Is it possible that this is sensor noise which would be "smoothed" away by JPEG compression, or by Colour/Luminance smoothing?

I've seen this in some shots with my 10D (EDIT: at ISO200) , lit via my (bounced) 420EX, but assumed it was sensor noise in lower-lit areas of the image.

Interested to see the answers from Canon though, as my 10D is still in the warranty period

Steve

Edited by Steve Knowles on Nov 17, 2004 at 04:18 PM GMT

stevei wrote:
I don't have a 20D, though I'm intrigued enough that I'll try it with my 300D this evening. I was assuming that if it were a common problem, then a lot of other people would have seen it too. On further reflection, though, I suppose there won't be that many people using the internal flash on long exposures, so it might be a common problem that the vast majority of people never experience.

Yeah, this is my hunch - this is a very uncommon use, I suspect. But I think it will prove to be a good example of a general 20D issue. You see the same sort of banding in other situations, but it isn't as noticeable. Hopefully having something that is so reproducable will help Canon nail the problem for future cameras.

stevei wrote:
I was also working on the basis that getting a replacement camera from the retailer is very little hassle, but I suppose if you've bought it over the internet then swapping it might not be so easy.

Yeah, that makes more sense. Unfortunately, I've found that even with smaller local retailers the arrangement with Canon is not such that they can simply swap out cameras with any confidence that Canon will credit them with a new camera. I know that if I went back to my dealer they would send the camera to Canon and I would be stuck with shooting with my Rebel for a couple weeks again. Until someone is able to prove this is not a general 20D problem, I'll be hanging on to mine.

Hopefully I'm wrong and the only reason I'm seeing it is because my 20D suffers the same electrical "fault" that jdee's does - but I'm skeptical, as always.

Steve Knowles wrote:
Are you shooting in RAW or not?

Is it possible that this is sensor noise which would be "smoothed" away by JPEG compression, or by Colour/Luminance smoothing?

jdee's images were out-of-cam JPEGs, mine are DPP-processed RAW files. This banding is so well-defined that noise-reduction does not reduce it by any acceptable agree. Others will disagree with that assessment - but for the way I think to noise-reduce photos, there's really no workaround.

Also, again - this issue is NOT limited to high-ISO work. jdee's samples were shot at ISO 200, mine were shot at ISO 100.

I'm not an electrical expert but my question would be is there a chance its picking up RF Interference and the 20D is less well shielded?

Was there any radio transmitting equipment, cell phone etc near to the camera when taking the shot. Sam's shot shows computers were they on when the shot was taken? - I know my small TV has horizontal bands and is unusable on some channels when my computer is on.

Will try some test shots with 20D and report back.

Jan

JVasek wrote:
I'm not an electrical expert but my question would be is there a chance its picking up RF Interference and the 20D is less well shielded?

Jan, I highly doubt this has anything to do with it. In this case you can isolate the issue down to the Flash by using it or not. In the problem I see, shots will be different shot-to-shot with identical settings - the key factor being whether the lens is hunting or not at the time the shot is take with AI Servo. In my case, I can make the problem appear or disappear by switching between AI Servo and One Shot mode.

Needless to say, if the 20D had problems with RF interference, you'd see a lot more people up in arms.

Sam as you said RFI unlikely - I've tried some 30 second shots held my phone (yes it was on) next to the camera no obvious difference.

I've tried ISO 3200 and 1600, long exposure with flash and without etc AI servo and lens hunting, and fortunately cannot get any banding as strong as you guys have shown, only feint banding at 3200 that cleaned up with noise reduction software.

I may try some more tests later with different lenses as well, sorry I have not been able to help.

Jan

I did call Canon and even sent them a link to a file and they said that they couldn't get the image. I would call them or email them (I think that is what I did) and have them look at the image full-size out of the camera and see what they say. I am going to do the same again and see what they reply with this time.

I've tried with a couple of lenses and I used the settings as noted in Sam Bennett's other thread. link to thread

I can confirm that I can create the same/similar banding at 100 ISO on my 20D using the internal flash, the banding in my case remains when I use AI-Servo or One shot and is most visible in darker areas of the shot. The shot was clean when I did not use the flash.

I've tried the same settings with a 550EX no banding. I also repeated this with the 10D just in case again no banding.

Hope this helps.

Jan