nhabedi Posted December 16, 2009 Share #1 Posted December 16, 2009 Advertisement (gone after registration) First of all, I have to admit that I followed all the discussion about high ISO banding only half-heartedly, mostly because up until I now I rarely used my M8.2 at ISO 640 and never at higher values. My apologies if I'm asking questions which have been answered before. I searched a bit in the archives of the forum, but the search function is kind of frustrating and I found contradictory reports about the issues I'm going to address. Anyway, a few days ago I thought I should try my luck at ISO 1250 and ISO 2500. One of the reasons was that I had seen some pretty nice high ISO shots from the M8 on this form recently. I should probably also mention that I'm only interested in black-and-white and I don't mind a bit of noise as long as I can convince myself that it has a grain-like quality. This posting certainly isn't about noise. Another disclaimer is that the examples I'm going to show are intentionally under-exposed by one or two stops. I've seen advice here to over-expose and then to darken the picture in post to get back the mood, but IMHO that's not the point of using high ISO in the first place. (And there might be no "reserve" left to over-expose. Maybe the lens is already full open and the shutter time so slow that you can barely hand-hold the camera.) I've also seen people claim that you have to really nail the exposure at ISO 1250 or higher, but I might expose for, say, a face which is sidelit by a bedside lamp and I still want the background to look OK although it'll be under-exposed. OK, issue #1 is that I sometimes see horizontal "bands" or "shadows" in the image. Here's one picture (at ISO 1250) where I see this. This is an out-of-camera JPEG, but the DNG has the same problem. The bands are in the upper left quarter of the picture, on the wall which is light grey in real life. Here's a 100% crop: Is this the well-known high-ISO banding, or is this something else? I've read here that there'll be a firmware fix for this, but it isn't there yet, or did I miss it? Issue #2 are vertical stripes with a different behaviour. First, here's a 100% crop from an out-of-camera JPG of a piece of wall at 2500: This is fine so far. It's again under-exposed and there's noise, but that's fine with me. But here's exactly the same photo from the DNG loaded into Lightroom and then exported: Notice the vertical stripe in the left third of the image? This is reproducible and the funny thing is that these stripes are always only in the DNGs, not in the JPGs. It also happens at ISO 1250. Is this also a known problem (maybe the same) or do I have to send my camera to Solms? If all these are old hats and fixes via firmware are announced, I don't have a problem waiting for this, just checking. Last question: What's the word on banding on the M9? Have folks seen similar problems there or is this issue gone for good? Thanks, Edi. Quote Link to post Share on other sites More sharing options...
Advertisement Posted December 16, 2009 Posted December 16, 2009 Hi nhabedi, Take a look here M8/M9 high ISO banding questions. I'm sure you'll find what you were looking for!
jaapv Posted December 16, 2009 Share #2 Posted December 16, 2009 (edited) If you underepose severely and push it past the limits of sensible photography you will get a kind of grid-like banding like a Scottish tartan. Any sensor will produce some kind of artefact under those circumstances. In general, you will always need to expose exactly in low light-high ISO situations to get the best out of any camera. Edited December 16, 2009 by jaapv Quote Link to post Share on other sites More sharing options...
dpattinson Posted December 16, 2009 Share #3 Posted December 16, 2009 I get banding in some shots - seems a little random in that they are not always the worst exposed ones and the banding can be quite evident without any post-processing exposure adjustment. Someone suggested that it gets worse if the sensor has been operating at high iso for a while due to retained heat. Recently at 2500iso I've started to get entire horizontal segments of the image with a different midtone density than the rest of the image. The area is always either at the top or bottom of the frame and is delineated with a very distinct edge. I can 'fix' it by masking and adjusting levels in post proc fairly easily, but it's nasty if the edge runs through important detail. This one is an example: london_2009 - sam on Flickr - Photo Sharing! I've done some work on it - but you can still detect the edge running through at the bottom, just at the level of the top of the tie knot. It was much more obvious before I worked it over. Quote Link to post Share on other sites More sharing options...
nhabedi Posted December 16, 2009 Author Share #4 Posted December 16, 2009 If you underepose severely and push it past the limits of sensible photography... I explained in detail why I think I didn't do that. Did you actually read that? I also did some comparison shots with my G1 and didn't see any bands or stripes there. I just asked some questions, there's no reason to go into defense mode. Quote Link to post Share on other sites More sharing options...
nhabedi Posted December 16, 2009 Author Share #5 Posted December 16, 2009 This one is an example: london_2009 - sam on Flickr - Photo Sharing! I've done some work on it - but you can still detect the edge running through at the bottom, just at the level of the top of the tie knot. It was much more obvious before I worked it over. Yes, that could be the same as my problem #1. So, is this the one Leica has promised a firmware fix for? Quote Link to post Share on other sites More sharing options...
dpattinson Posted December 16, 2009 Share #6 Posted December 16, 2009 I would say - that at 1250 or above, it doesn't pay to have any important detail in the shadow areas of the image. It's pretty much like shooting 3200 film, it works much better if the lighting is contrasty and/or you can let the shadows darken a lot. Quote Link to post Share on other sites More sharing options...
dpattinson Posted December 16, 2009 Share #7 Posted December 16, 2009 (edited) Advertisement (gone after registration) Yes, that could be the same as my problem #1. So, is this the one Leica has promised a firmware fix for? I can't really see anything in your first image that looks as bad as what I see in my unprocessed jpeg or dng. In my case the entire bottom 10% of the image ended up significantly lighter than the rest of it. I don't pay any attention to what people say Leica has promised fixes for, since they never release fix notes of any value - and they never publish any list of known issues... I haven't even bothered to report this new issue - because I'm quite sure the response will be that it's all very surprising and I should send my camera in for 3 months and get it back with a note that says 'sensor replacement' or the like... sorry if that comes off as a bit pissy... but I develop software for a living and I think I know what industry best practice looks like in terms of handling defect reports and fixes. Edited December 16, 2009 by dpattinson Quote Link to post Share on other sites More sharing options...
jaapv Posted December 16, 2009 Share #8 Posted December 16, 2009 (edited) I explained in detail why I think I didn't do that. Did you actually read that? I also did some comparison shots with my G1 and didn't see any bands or stripes there. I just asked some questions, there's no reason to go into defense mode. Excuse me? You asked what type of banding under which circumstances on the M9. What is wrong with answering a question you ask??Where did I comment on your exposure? I told you how I expose to get which type of banding on my M9. Reading seems to be difficult... Edited December 16, 2009 by jaapv Quote Link to post Share on other sites More sharing options...
graeme_hutton Posted December 16, 2009 Share #9 Posted December 16, 2009 (edited) Re. M8 - This has been discussed at some length. Leica have promised a firmware fix in the near future. Best Graeme Edited December 16, 2009 by graeme_hutton Quote Link to post Share on other sites More sharing options...
nhabedi Posted December 16, 2009 Author Share #10 Posted December 16, 2009 Excuse me? You asked what type of banding under which circumstances on the M9. What is wrong with answering a question you ask?? Nothing. Except that you didn't answer any of my questions. I only saw sweeping statements about "severe underexposure" and "pushing past the limits of sensible photography." Where did I comment on your exposure? I don't know. Did someone say that you commented on my exposure? I told you how I expose to get which type of banding on my M9. Where please did you say that your answer was about the M9? Where please did you say how you expose to get which type of banding? Did you mean the sentence "In general, you will always need to expose exactly in low light-high ISO situations to get the best out of any camera"? Reading seems to be difficult... Yes, but I don't hold it against you. Must be your age... Quote Link to post Share on other sites More sharing options...
jaapv Posted December 16, 2009 Share #11 Posted December 16, 2009 Last question: What's the word on banding on the M9? Have folks seen similar problems there or is this issue gone for good?................ Quote Link to post Share on other sites More sharing options...
nhabedi Posted December 16, 2009 Author Share #12 Posted December 16, 2009 Re. M8 - This has been discussed at some length. Leica have promised a firmware fix in the near future. Yes, I heard about the promised firmware fix. My question was if both examples I provided are the same issue or if they're different (one is horizontal, one is vertical, one is kind of fuzzy while the other one is pretty pronounced, one can be seen in DNGs as well as in JPGs, the other one only occurs in DNGs). And if they are different, I'd of course like to know which one is the soon-to-be-fixed one and if the other one maybe is a problem with my particular camera. In case someone happens to know the answer, that is... Quote Link to post Share on other sites More sharing options...
nhabedi Posted December 16, 2009 Author Share #13 Posted December 16, 2009 ................ Ah, so it was a contorted way of saying that the M9 has the same problems? Thanks. Quote Link to post Share on other sites More sharing options...
adan Posted December 16, 2009 Share #14 Posted December 16, 2009 Important note: In digital, you really only have the equivalent of a single ISO, in the film sense. The base ISO (which is 160 for the M8/M9). When you shoot at higher ISOs, you are already "pushing" that "film" 1,2,3 or 4 stops, just as one could shoot Fuji 160 film at 640 or 2500 and then push by overdeveloping. So the instant you move the ISO off 160 - you are already "underexposing." I.E. it is not the same as just loading ISO 2500 film in the first place. Underexpose at 2500 by 1-2 stops and you are shooting at 5000 or 10,000 ISO. Anyway - to get to your questions: 1) horizontal banding in underexposed high-ISO digital images will never be "gone for good". Eventually, you can make any camera band, if you feed the sensor so little light that the background noise starts to dominate. Things will likely improve over the decades - but I'll guarantee you that I'll be able to make a Nikon D900s band at ISO 2,000,000 in 10 years or so. 2) Is the M9 banding ever so slightly less obvious than the M8's at ISO 2500. Yes, slightly. Not much. 3) As to the vertical stripe - there are several "instances" of this. One, that is consistent in only showing up at ISO 2500 and on the right side, and in many M9s, has been diagnosed as a read error and will likely be one of the fixes in the next firmware edition. Other more randomly located vertical lines that show up at several (or all) ISOs seem to be specific sensor bugs and (I think) require sensor replacement. I think yours may be the second type, by location - but wait for the FW upgrade to see. You may avoid unnecessary M9 surgery. 4) There does seem to be some differences in what shows up in jpegs vs. dngs. I've seen hot single pixels (usually red, sometimes blue) that are very obvious in jpegs but invisible in the corresponding dng (shot as dng + jpeg). Quote Link to post Share on other sites More sharing options...
nhabedi Posted December 16, 2009 Author Share #15 Posted December 16, 2009 Andy, thanks for the detailed reply. I'll wait for the firmware fix before I send my camera to Solms, it's not a big issue for me right now and it's older than one year anyway. (It's an M8.2, not an M9.) What concerns be about the "underexposure problem" is that this will inevitably happen in low-light situations. I've shown some extreme examples here, but I noticed this in real-world cases where the important part of the picture was properly exposed and the background was simply pretty dark (because it wasn't lit). I wouldn't mind letting the shadows darken there as David suggested, but even if I do that in post, I have to manually retouch the banding. As you said, with film I simply wouldn't worry about this (or about "the limits of sensible photography" for that matter). Quote Link to post Share on other sites More sharing options...
theendlesshouse Posted December 16, 2009 Share #16 Posted December 16, 2009 I have never had banding on any of my digital cameras at any level of iso or exposure which would lead me to the conclusion that this was a fault! Quote Link to post Share on other sites More sharing options...
nhabedi Posted December 22, 2009 Author Share #17 Posted December 22, 2009 Update: I've sent some of my sample photos to Leica and asked for advice. I received a reply by email today. Interestingly, they didn't talk about "the limits of sensible photography" or even underexposure. Instead, they acknowledged this to be a known problem of the M8. The explanation is roughly that the M8 sensor only receives a very low signal at high ISO values and thus the process of creating the image is sensitive to (the effects of) simultaneous processes like for example shutter recock. They're working on electronically de-coupling these processes and hope to have a firmware update available in the first quarter of 2010. Once again, kudos to Leica for always being open about such issues and for always providing detailed replies to their customers. I have to admit, though, that I would have hoped that this problem was fixed by now given that the M8 is more than three years old. I didn't ask about the M9. But I guess if they knew how to fix this in the M9 we'd have a fix for the M8 already. Quote Link to post Share on other sites More sharing options...
dpattinson Posted December 22, 2009 Share #18 Posted December 22, 2009 Update: I've sent some of my sample photos to Leica and asked for advice. I received a reply by email today. Interestingly, they didn't talk about "the limits of sensible photography" or even underexposure. Instead, they acknowledged this to be a known problem of the M8. The explanation is roughly that the M8 sensor only receives a very low signal at high ISO values and thus the process of creating the image is sensitive to (the effects of) simultaneous processes like for example shutter recock. They're working on electronically de-coupling these processes and hope to have a firmware update available in the first quarter of 2010. Once again, kudos to Leica for always being open about such issues and for always providing detailed replies to their customers. I have to admit, though, that I would have hoped that this problem was fixed by now given that the M8 is more than three years old. I didn't ask about the M9. But I guess if they knew how to fix this in the M9 we'd have a fix for the M8 already. OK - that is very good news indeed Quote Link to post Share on other sites More sharing options...
Nicoleica Posted December 22, 2009 Share #19 Posted December 22, 2009 Update: I've sent some of my sample photos to Leica and asked for advice. I received a reply by email today. Interestingly, they didn't talk about "the limits of sensible photography" or even underexposure. Instead, they acknowledged this to be a known problem of the M8. The explanation is roughly that the M8 sensor only receives a very low signal at high ISO values and thus the process of creating the image is sensitive to (the effects of) simultaneous processes like for example shutter recock. They're working on electronically de-coupling these processes and hope to have a firmware update available in the first quarter of 2010. Once again, kudos to Leica for always being open about such issues and for always providing detailed replies to their customers. I have to admit, though, that I would have hoped that this problem was fixed by now given that the M8 is more than three years old. I didn't ask about the M9. But I guess if they knew how to fix this in the M9 we'd have a fix for the M8 already. Interesting. Thanks for posting this. I wonder if it would be worth conducting an experiment to see if the banding is reduced if discrete mode is used? This could reduce power fluctuations from the re-cocking procedure if one waited until the red light stopped flashing before releasing the shutter. Quote Link to post Share on other sites More sharing options...
dpattinson Posted December 22, 2009 Share #20 Posted December 22, 2009 Actually - I suspect discreet mode might be a culprit in this related issue. As I recall, discreet mode draws additional power to keep the shutter from winding on? I seem to recall it has a negative effect on battery life from the original analysis when it was released. I may try to reproduce my issue tonight and see if turning off discreet mode has any effect. Quote Link to post Share on other sites More sharing options...
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.