nhabedi Posted October 14, 2009 Share #41 Posted October 14, 2009 Advertisement (gone after registration) Suppose the pixels you are comparing give you similar results on a per-pixel basis, i.e. signal-to-noise ratio and dynamic range are similar. Now when you’ve got more of those equally good pixels and the increased resolution is more than either your printer or your eyes (or both) can handle, then some of those pixels will get averaged (by the printer or your eyes), effectively reducing noise and thereby increasing dynamic range. So more pixels give you increased resolution, up to the point where still more resolution in the image would be wasted. From that point onwards, more pixels increase the signal-to-noise ratio and dynamic range. In most cases these days we are dealing with more pixels crammed onto the same sensor size; in those cases, the signal-to-noise ratio and dynamic range drop, measured on a per-pixel basis, and the increased resolution cannot quite make up for that. But in the case of the M9 versus the M8 we have rather similar (although slightly improved) sensor pixels, only more of those, resulting in a net gain. Yes, I fully agree with you. I was talking about the comparison of the D3x to the M9 where the D3x effectively "crams" more pixels into the same ("full frame") sensor area. Maybe the D3x pixels aren't only equally good but actually better - the question still remains whether you could notice a difference in prints up to A2. (I'm not saying you can't. I'm just saying that a "test" like that of Mr. Puts doesn't answer this question at all, so it's useless to me.) As to M8 vs. M9 comparisons I also agree (again, without having a M9 myself). Those who claim that the M9 is identical to a M8 in image quality must surely refer to the pixel level, i.e. they're not comparing prints but they're pixel-peeping. (I'm old-fashioned and I still think that cameras, even digital ones, should be used to take pictures and not to create pixels...) Link to post Share on other sites More sharing options...
Advertisement Posted October 14, 2009 Posted October 14, 2009 Hi nhabedi, Take a look here Erwin Puts: The Leica M9: part 5: M8/9 noise and dynamic range. I'm sure you'll find what you were looking for!
nhabedi Posted October 14, 2009 Share #42 Posted October 14, 2009 Now when you’ve got more of those equally good pixels and the increased resolution is more than either your printer or your eyes (or both) can handle, then some of those pixels will get averaged (by the printer or your eyes), effectively reducing noise and thereby increasing dynamic range. Oh, wait, there's one point I don't get here. I can understand how averaging more pixels can effectively reduce noise. But how can it increase dynamic range? Link to post Share on other sites More sharing options...
Angelos Viskadourakis Posted October 14, 2009 Share #43 Posted October 14, 2009 In most cases these days we are dealing with more pixels crammed onto the same sensor size; in those cases, the signal-to-noise ratio and dynamic range drop, measured on a per-pixel basis, and the increased resolution cannot quite make up for that. But in the case of the M9 versus the M8 we have rather similar (although slightly improved) sensor pixels, only more of those, resulting in a net gain. the S/N is depedent on photosite size and overal quality of wiring and data transfer consequently cannot be improved further using the same sensor and i cannot see anywhere or anybody to claim that the pixels in M9 are slightly improved and the huge issue is that in order to correct the vigneting in the larger sensor-added material in both sides-a software is used that lightens up the darker corners but increasing the noise,i dont know about others but i like much better darker sides than noisier sides that give a different texture in the image,ofcourse any M BODY is a joy to work with ,the point is for LEICA sales not to force their excellent engineers to introduce something hypothetically improved when is not so. Link to post Share on other sites More sharing options...
delander † Posted October 14, 2009 Share #44 Posted October 14, 2009 I'm really depressed, reading all this it seems I've bought an extremely expensive lemon:(. Still the colour seems good;). Jeff Welcome, dear visitor! As registered member you'd see an image here… Simply register for free here – We are always happy to welcome new members! Link to post Share on other sites Simply register for free here – We are always happy to welcome new members! ' data-webShareUrl='https://www.l-camera-forum.com/topic/100096-erwin-puts-the-leica-m9-part-5-m89-noise-and-dynamic-range/?do=findComment&comment=1074629'>More sharing options...
t024484 Posted October 14, 2009 Share #45 Posted October 14, 2009 Erwin shows us the results of something that took quite some energy to achieve. Someone suggested that this is just for pixel peepers or technical nerds. Let everybody decide for himself how he wants to enjoy his life. Without pixel peepers and technicians there would never have been a M9. Now for the tests itself. The question is, how should the outcome of this all be read. Is this test validated by comparing it against the outcome of other measuring principles? Are Jpegs from Leica camera´s suited for this test, I think not. A M8.2 tests in Camera Magazine 120 (a Dutch Magazine) also performed with Imatest, showed a Dynamic Range of 11,6 stops at ISO 160 based upon RAW images. What a difference that is to this test. So the least one can say is that Imatest can produce different figures depending on who is using it. A simple calculation that one can make for the M8, based upon facts in Kodak's documentation shows: Full exposure for the M8 Sensor corresponds to 60.000 electrons. Shot noise at that point is Sqrt(60.000) = 245 electrons. If shot noise is the only noise generated, S/N= 20* Log ( 60.000 / 245) = 47.8 dB. Real world S/N can only be worse because of other types of noise, but never be better. In Mr Puts graphs for the M8 at ISO 160, the lower of the three graphs shows a S/N of roughly 55 dB at full exposure, more than 7dB over the top. The same calculation for ISO 2500 gives a theoretical maximum S/N of 35.8 dB. In this article it is 45 dB, almost 10dB more than nature allows! So the figures for “Log Exposure” at 0 on the horizontal axis are much too high, but …. given the fact that shot noise is exactly the Sqrt ot the number of electrons, it follows that if the amount of electrons goes down with 40 dB (or -2 at the horizontal axis) , the shot noise goes down with 20 dB. Now have a look at the M8 graph at ISO 2500. From 0 to -2. the noise goes down from 45 to 0dB, which is a difference of 45 dB instead of 20 dB, a slope that is far too steep and therefore fully unrealistic. Also at ISO 160 the slope is too steep, although not as steep as at ISO 2500, because of that projecting an S/N that is far too low. So for some reason both the starting points, and also the slopes seem to be wrong in these graphs, leading to fully wrong results, and they do not correspond in any way with the many tests that I have done. So my conclusion can only be that the outcome of this test is not in any way connected to real world. Hans Link to post Share on other sites More sharing options...
t024484 Posted October 14, 2009 Share #46 Posted October 14, 2009 Oh, wait, there's one point I don't get here. I can understand how averaging more pixels can effectively reduce noise. But how can it increase dynamic range? If the noise is reduced, and the main signal is still the same, the S/N or dynamic range is increased. Hans Link to post Share on other sites More sharing options...
mjh Posted October 14, 2009 Share #47 Posted October 14, 2009 Advertisement (gone after registration) Oh, wait, there's one point I don't get here. I can understand how averaging more pixels can effectively reduce noise. But how can it increase dynamic range? Dynamic range is limited by the pixel’s full-well capacity on one side and the noise level on the other. Clipping due to the accumulated charge exceeding the full-well capacity is a hard limit, but you can do something about noise. Starting with the full-well capacity, each f-stop of dynamic range corresponds to halving the number of electrons accumulated. Assuming that the sensor of the M9 is sufficiently similar to that of the M8, we would have a maximum of about 60,000 electrons. 14 f-stops down we are left with 3.7 electrons, corresponding to a signal-to-noise ratio of 1.9 due to shot noise alone – that’s 2 parts signal and one part noise. Obviously it goes downhill from those values if we increase the ISO setting beyond the sensor’s native sensitivity. Doubling the ISO sensitivity increases noise levels by about 40 percent, amounting to a 1/2 f-stop loss in dynamic range. Conversely, any improvement in the signal-to-noise ratio, even by simply averaging neighboring pixels, increases dynamic range. Link to post Share on other sites More sharing options...
rosuna Posted October 14, 2009 Share #48 Posted October 14, 2009 So for some reason both the starting points, and also the slopes seem to be wrong in these graphs, leading to fully wrong results, and they do not correspond in any way with the many tests that I have done. Hans, What are your conclusions about the camera? How much differ from those of Puts? Link to post Share on other sites More sharing options...
mjh Posted October 14, 2009 Share #49 Posted October 14, 2009 the S/N is depedent on photosite size and overal quality of wiring and data transfer consequently cannot be improved further using the same sensor and i cannot see anywhere or anybody to claim that the pixels in M9 are slightly improved and the huge issue is that in order to correct the vigneting in the larger sensor-added material in both sides-a software is used that lightens up the darker corners but increasing the noise,i dont know about others but i like much better darker sides than noisier sides that give a different texture in the image,ofcourse any M BODY is a joy to work with ,the point is for LEICA sales not to force their excellent engineers to introduce something hypothetically improved when is not so. That’s some awfully long sentence … Anyway, one factor affecting noise is the difference in transmission between the red, green, and blue color filters in front of the pixels, and that is an area where the sensor is said to be improved. Lacking a spec sheet for the KAF-18500 the effect is impossible to quantify, but it is not like Kodak just designed a KAF-10500 Mark II, only with more pixels. Link to post Share on other sites More sharing options...
nhabedi Posted October 14, 2009 Share #50 Posted October 14, 2009 Dynamic range is limited by the pixel’s full-well capacity on one side and the noise level on the other. Clipping due to the accumulated charge exceeding the full-well capacity is a hard limit, but you can do something about noise. Starting with the full-well capacity, each f-stop of dynamic range corresponds to halving the number of electrons accumulated. Assuming that the sensor of the M9 is sufficiently similar to that of the M8, we would have a maximum of about 60,000 electrons. 14 f-stops down we are left with 3.7 electrons, corresponding to a signal-to-noise ratio of 1.9 due to shot noise alone – that’s 2 parts signal and one part noise. Obviously it goes downhill from those values if we increase the ISO setting beyond the sensor’s native sensitivity. Doubling the ISO sensitivity increases noise levels by about 40 percent, amounting to a 1/2 f-stop loss in dynamic range. Conversely, any improvement in the signal-to-noise ratio, even by simply averaging neighboring pixels, increases dynamic range. OK, that makes sense now. Thanks for the explanation. Link to post Share on other sites More sharing options...
Angelos Viskadourakis Posted October 14, 2009 Share #51 Posted October 14, 2009 my apologies for the long sentence but i like them like that -i also like panoramas and long stiched photos. my point is that hypothetically we need a better next model to M8 is this advancement the M9?????? that is all we can expect????? , plus i'm wondering what will be the form of the M8 UPGRADE as is promised in public many times from LEICA. Link to post Share on other sites More sharing options...
stunsworth Posted October 14, 2009 Share #52 Posted October 14, 2009 ...plus i'm wondering what will be the form of the M8 UPGRADE as is promised in public many times from LEICA. There's won't be any more upgrades to the M8, it's been replaced by a different model. Link to post Share on other sites More sharing options...
georg Posted October 14, 2009 Share #53 Posted October 14, 2009 +2.5 stops DR with the 5D? Far superior DR with any current DSLR? Please, someone show me a real-world 1:1 comparison (of course NOT in-camera JPGs) that show this advantage! Link to post Share on other sites More sharing options...
mjh Posted October 14, 2009 Share #54 Posted October 14, 2009 my point is that hypothetically we need a better next model to M8 is this advancement the M9?????? that is all we can expect????? Well, not so hypothetically customers are standing in line for an M9. i'm wondering what will be the form of the M8 UPGRADE as is promised in public many times from LEICA. There will be no hardware upgrade. I expect some bug-fixes and maybe some improvements with another firmware update, but that’s about it. Link to post Share on other sites More sharing options...
rosuna Posted October 14, 2009 Share #55 Posted October 14, 2009 Well, not so hypothetically customers are standing in line for an M9. Well, we don't know how many M9s will be sold. Leica plans to ship 6,000 units until January 2010. That is a fast production pace. The stock of potential customers will be depleted soon. The problem is how to sell a camera of that price with the noise and DR performance of the M8 4 years later (2006-2010), 6 years if we consider the DMR (with a sensor very similar to that of the M8). The full format achievement is enough to catch many M8 users, and some new ones (coming from the film M cameras, or new to Leica), but in order to get a sustained demand for the camera it has to be in line with the IQ of the similarly priced DSLR cameras of the moment. The translation of some of the S2's advances to the M line can't wait 2 or 3 years, that is sure. I am anxious to see how the S2 performs, compared to the M8/M9, in terms of noise and dynamic range (considering the smaller pixels). Link to post Share on other sites More sharing options...
mjh Posted October 14, 2009 Share #56 Posted October 14, 2009 The problem is how to sell a camera of that price with the noise and DR performance of the M8 4 years later (2006-2010), 6 years if we consider the DMR (with a sensor very similar to that of the M8). The full format achievement is enough to catch many M8 users, and some new ones (coming from the film M cameras, or new to Leica), but in order to get a sustained demand for the camera it has to be in line with the IQ of the similarly priced DSLR cameras of the moment. The translation of some of the S2's advances to the M line can't wait 2 or 3 years, that is sure. Contrary to popular belief there weren’t any real breakthroughs in sensor technology these past years. There was some gradual improvement to be sure, but most advances have come from software. If it is lower noise levels and more dynamic range that you want, then get a sensor with bigger pixels; that’s the secret, basically. Most anything else is improved noise reduction. So if you keep your M9 (or an M8 for that matter) during the coming years, but make it a point of always getting the best raw converter that’s out there, your camera will probably stay competitive. There’s much less potential for an improvement of image processing in-camera, though, as it is limited by the processor(s) used. Link to post Share on other sites More sharing options...
harryzet Posted October 14, 2009 Share #57 Posted October 14, 2009 who cares what this bribed guy writes? Link to post Share on other sites More sharing options...
rosuna Posted October 14, 2009 Share #58 Posted October 14, 2009 Contrary to popular belief there weren’t any real breakthroughs in sensor technology these past years. There was some gradual improvement to be sure, but most advances have come from software. If it is lower noise levels and more dynamic range that you want, then get a sensor with bigger pixels; that’s the secret, basically. Most anything else is improved noise reduction. So if you keep your M9 (or an M8 for that matter) during the coming years, but make it a point of always getting the best raw converter that’s out there, your camera will probably stay competitive. There’s much less potential for an improvement of image processing in-camera, though, as it is limited by the processor(s) used. Do more channels at lower frequencies for extracting the signal have a substantial impact in noise and DR? I am referring to CCD sensors. This is the most important "improvement" of the S2 sensor compared to that of the DMR/M8/M9. CMOSes have evolved during the past years, or it seems so. The total number of pixels have increased by much, but also bit depth (from 12 to 14 bits) and noise at high ISOs has improved as well. It may be due to in camera noise reduction, but it works. I suppose there are internal "software" adjustments, but also electronic improvements in the sensor itself. Anyway, even assuming that Puts' methodology is a bit strange, the broad picture from the comparison seems to be correct: CMOSes, whatever the techniques applied, have much better performance at high ISOs, even with smaller pixels... and postprocessing on CCDs files cannot reduce this advantage. You cannot recover the DR lost when you push the ISO above 640. Maybe the solution for the M line isn't a reduced version of the S2 sensor, but a CMOS sensor... I don't know, but the gap is growing and growing... Link to post Share on other sites More sharing options...
UliWer Posted October 14, 2009 Share #59 Posted October 14, 2009 @harryzet What you write, is just an insult. Many others in this thread have given arguments for their perception of the topic. Your insult has nothing to do with this. Link to post Share on other sites More sharing options...
luigi bertolotti Posted October 14, 2009 Share #60 Posted October 14, 2009 @harryzet What you write, is just an insult. Many others in this thread have given arguments for their perception of the topic. Your insult has nothing to do with this. Endorse your observation : not fair to insult a men who puts (...sorry...) his works freely available and open to everyone's for discussion. Link to post Share on other sites More sharing options...
Recommended Posts
Archived
This topic is now archived and is closed to further replies.