diogenis Posted October 21, 2009 Share #21 Posted October 21, 2009 Advertisement (gone after registration) Hans' explanations are quite elaborate, so I guess it is easily missed -- but he came to the conclusion that the 1 stop improvement is a result of both lower sensor noise *and* more pixels: Yea, you are right I did missed that So, one full stop advantage in noise for the M9 and the only other way of improving it without IQ degradation is by using a second parallel amp, or by lowering read time of the single one. Can this be done through firmware? Link to post Share on other sites More sharing options...
Advertisement Posted October 21, 2009 Posted October 21, 2009 Hi diogenis, Take a look here The M9 has indeed one stop advantage over the M8. I'm sure you'll find what you were looking for!
Leicaiste Posted October 21, 2009 Share #22 Posted October 21, 2009 I didn't want the FF for more field of view with the same lens, but to be able to use the lenses for what they were designed for. . Isn't that the same thing? Guido, When I used a 28mm on the M8 it is not because I wanted a 28mm field of view, but because I wanted to mimic a 35mm (ok 37mm) on the M6/M7. The only advantage was digital versus film. Now, I prefer to use a 35/2 on the M9 because the lens is smaller, cheaper, its field slightly wider, the frames more comfortable and the sensor offers more resolution with less noise. The same almost apply to the 24/1,4 (= 32) on the M8 versus the 35/1,4 on the M9. Etc... Again, those who started M photography with the M8 may not see my point. And they may prefer the 50/1,4, 75/2 and 90/2 on the M8. I certainly don't, save maybe the 50/1,4. If I had to start from scratch today, I will prefer to do it with an M9 + 28/2, 35/1,4 or f2, 50/1,4, 75/2 and maybe 90/2 than with an M8.2 + 21/2,8 (28), 24/1,4 (32) or 28/2 (37), 35/1,4 (46), 50/1,4 (66) and maybe 75/2 (100). Lucien Link to post Share on other sites More sharing options...
t024484 Posted October 21, 2009 Author Share #23 Posted October 21, 2009 Thanks a lot , these are very interesting findings, it is not so clear to me why the new test gives much more solid results, so, I do not really understand what was wrong in the first test. Anyway, I am willing to believe the second new series of tests which confirms Leica's statement that the M9 has almost one stop advantage in noise level over the M8. On the other hand, it puts the question, why other reviewer for example Erwin Puts come to the opposite conclusion. Do you have an explanation? Thomas I don't know what first test you are referring to. My own calculation of the RN of the M8 was too low, because it is quite dufficult to isolate, as i explained above. Now based on many measurements, I have corected the initial figure. Erwin Puts has used Imatest, and used Jpeg's for that. Imatest is a rather indirect test that can produce quite different results. As an example, look below. Two Imatests for the same M8 camera, both at ISO 160, both using a Jpeg. The dynamic Range at quality level 0.1 differs 1.5 Stops, and at 0.25 even 1,9 Sops, yet the camera is still the same. 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! The same Imatest performed on a TIF, generated from RAW, shows an increase in dynamic range from 9.09 to 11.5. A bit high to my taste, because I calculated a DR of 10.5 stops for the M8. So different people can come to different Imatest results, meaning that you must be carefull in interpreting the results. My tests on the other hand are very straightforward, but time consuming. I measured the real signal, the noise, etc, etc. Hans Link to post Share on other sites Simply register for free here – We are always happy to welcome new members! The same Imatest performed on a TIF, generated from RAW, shows an increase in dynamic range from 9.09 to 11.5. A bit high to my taste, because I calculated a DR of 10.5 stops for the M8. So different people can come to different Imatest results, meaning that you must be carefull in interpreting the results. My tests on the other hand are very straightforward, but time consuming. I measured the real signal, the noise, etc, etc. Hans ' data-webShareUrl='https://www.l-camera-forum.com/topic/100892-the-m9-has-indeed-one-stop-advantage-over-the-m8/?do=findComment&comment=1083617'>More sharing options...
t024484 Posted October 21, 2009 Author Share #24 Posted October 21, 2009 This is very interesting , Hans. As Sean said, it matches his observations, and it also matches what S2 project manager Stephan Shulz said in response to question #2: http://www.l-camera-forum.com/leica-forum/leica-s2-forum/95115-press-release-s2-technical-specs-7.html#post1002971 I wonder if it would be technically feasible to have a new option in future firmware for the M8/M9, and if so, would Leica be interested? Myself, I hardly ever shoot many fps, rather, I do one shot here and another there, so if I could set my camera to slow/low-noise I would probably have that setting on at all times. Could Leica work more with Kodak on read noise for the future M10? Regards I do not think that the M8/M9 is equiped with a programmable clock. So a firmware update to reduce the clock would be out of the question. A future M10 could very well have less noise. There are several ways to go, to name a few: Paralleling amplifiers wil reduce the Noise with the SQRT. So 4 amplifiers in parallel will reduce read noise by a factor 2. Another more complex option is to use a number of slower amps, where a switch is used to divide pixel charges over the amps, and a switch behind the amps to collect the results for the A/D conversion. Hans Link to post Share on other sites More sharing options...
sean_reid Posted October 21, 2009 Share #25 Posted October 21, 2009 Hans' explanations are quite elaborate, so I guess it is easily missed -- but he came to the conclusion that the 1 stop improvement is a result of both lower sensor noise *and* more pixels: That's my understanding as well. Of course any of these results may vary a bit depending on subject lighting (colder or warmer) etc. but what I've seen, and written about in my review, were some differences when comparing files at native size but a much more noticeable difference when comparing the two files sized to match. If you reread the review you'll note that. A friend of mine who beta tested the M9 and who owns it now feels that the difference is a bit more than a stop but those sorts of distinction one could debate for a long time, depending on the exact test pictures being compared. In any case, I think describing the difference as about a stop is accurate. After my ISO tests were published, an engineer at Leica wrote to confirm that they were consistent with what Leica's own tests of the production model were showing. And now Hans has used different methods to look at the question and come to about the same conclusion as well. Cheers, Sean Link to post Share on other sites More sharing options...
sean_reid Posted October 21, 2009 Share #26 Posted October 21, 2009 Sean, if I can remember, your observations were also 1 stop difference but the reasoning was that this was exhibiting due to more information from the larger sensor printed on paper. However Hans with his research proved a real 1stop difference on sensor level. I believe that your reasoning still stands, so shouldn't the outcome be then 2 stops instead of one (better sensor + larger sensor) on printed photos???In short question is: the effect of noise diminishes when more pixels are involved or not? Hans can answer this but I don't think his tests suggest a one-stop difference at actual pixel level (without matching size). I believe they find about a stop difference with the files sized to match (various factors combined). Cheers, Sean Link to post Share on other sites More sharing options...
sean_reid Posted October 21, 2009 Share #27 Posted October 21, 2009 Advertisement (gone after registration) One thing to keep in mind when doing noise testing is that the subject lighting (warmer or cooler) will normally have an affect on noise levels because one would normally white balance the file in conversion from RAW. The warmer the test lighting, the more the conversion will need to draw upon the blue channel to balance the color. So there will always be some variation in test results depending upon the conditions under which the test pictures themselves are made. It's possible, of course, to look at the noise in a file that hasn't been white balanced but that wouldn't necessarily represent the way one would work with a file in actual use. Hans' results, from Tim's pictures, are consistent with what several others have reported using various methods. I'm curious, Tim and Hans, about the lighting used for the test pictures and the method used to set white balance for each one. I normally test under studio tungsten (quite warm) and set WB by sampling a WhiBal card. Thanks again for the work you've done. Cheers, Sean Link to post Share on other sites More sharing options...
sean_reid Posted October 21, 2009 Share #28 Posted October 21, 2009 Hi Sean, You arrived at the same conclusion about the final print, which I suppose is all that matters in the end. But there also seem to be some interesting disagreements (or perhaps differences if "disagreements" sounds you two are stuck in a debate, which clearly you're not). OP seems to be saying at even at base ISO (before you believed any software smoothing came into effect), there is better S/N *at the pixel level*. I might have missed recent updates, but my memory is that you saw no advantage at the pixel level and attributed all of the gain in S/N to downsampling the M9 and/or upsampling the M8. David Hi David, Take a look at the review again. One can see some differences in the files there even when they're at actual size (esp. at ISO 2500) but a much more noticeable difference when they're sized to match. As you say, there's no debate here as Hans, myself and several others have all come to about the same conclusions using different means. Cheers, Sean Link to post Share on other sites More sharing options...
sandymc Posted October 21, 2009 Share #29 Posted October 21, 2009 Imatest is a rather indirect test that can produce quite different results. The sad fact is, and I think Michael Hußmann has made this same point on other threads, is that what Imatest is really comparing is noise plus demosaicing plus JPEG processing; different processing, different answer. If you're really, really careful, you might be able to get comparable results from Imatest, but the JPEG converters on modern cameras and/or raw converters are so sophisticated that you really can't guarantee comparability, no matter how hard you try. Sandy Link to post Share on other sites More sharing options...
jaapv Posted October 21, 2009 Share #30 Posted October 21, 2009 Perhaps Leica would be better served sending you a camera to test Hans? rather than Herr Puts? Erwin is no Herr - that is reserved for German gentlemen. If you want to adress him in his own language, try " Meneer" Link to post Share on other sites More sharing options...
sean_reid Posted October 21, 2009 Share #31 Posted October 21, 2009 I'm curious, Tim and Hans, about the lighting used for the test pictures and the method used to set white balance for each one. Sean Forgive me, that should have read: "Chris and Hans" Cheers, Sean Link to post Share on other sites More sharing options...
Guest EarlBurrellPhoto Posted October 21, 2009 Share #32 Posted October 21, 2009 I don't mean to demean anyone for taking the time to test out the M9 and prove that it is one stop better with NR than the M8, but the truth is, it's still far behind the benchmark. If high-ISO noise is a deal-maker/breaker for you, the M9 even if 1 stop better than the M8, is not your best choice. The reasons to choose an M9 or M8 are the same as they were when there were only film cameras: fabulous glass, small footprint, and rangefinder viewing. Link to post Share on other sites More sharing options...
t024484 Posted October 21, 2009 Author Share #33 Posted October 21, 2009 Hans with his research proved a real 1stop difference on sensor level. I believe that your reasoning still stands, so shouldn't the outcome be then 2 stops instead of one (better sensor + larger sensor) on printed photos???In short question is: the effect of noise diminishes when more pixels are involved or not? Lower front end Noise brings roughly 0.7 Stop improvement, more available pixels when printing 0.4 Stop, together 1.1 Stop. The Printing effect of more pixels goes on all the way up with the Sqrt as long as the pixel size remains the same. A 48 *72 mm 72 Mpixel sensor would give another factor 2 or 1 extra stop in Noise reduction when still printing the same composition on the same paper dimension. A 72 Mpixel sensor of 24 * 36 mm would not bring any printing advantage over the current 18 Mpixel sensor. Hans Link to post Share on other sites More sharing options...
diogenis Posted October 21, 2009 Share #34 Posted October 21, 2009 Thanks Hans, great work by the way. Link to post Share on other sites More sharing options...
sean_reid Posted October 21, 2009 Share #35 Posted October 21, 2009 I don't mean to demean anyone for taking the time to test out the M9 and prove that it is one stop better with NR than the M8, but the truth is, it's still far behind the benchmark. If high-ISO noise is a deal-maker/breaker for you, the M9 even if 1 stop better than the M8, is not your best choice. The reasons to choose an M9 or M8 are the same as they were when there were only film cameras: fabulous glass, small footprint, and rangefinder viewing. On the other hand, though, photographers who do work at high ISO (and who want to work with a DRF) will find it useful to know that they're gaining a stop with the M9 as compared to the M8. It's true that one gets better high ISO performance with certain Canon and Nikon DSLRs for example but, of course, they are different kinds of cameras with the usual SLR pros and cons. Cheers, Sean Link to post Share on other sites More sharing options...
adan Posted October 21, 2009 Share #36 Posted October 21, 2009 The reasons to choose an M9 or M8 are the same as they were when there were only film cameras: fabulous glass, small footprint, and rangefinder viewing. True, and I still choose fabulous glass, small footprint, and rangefinder viewing even when the light gets dim - well, almost fabulous: I use pre-ASPH lenses (but 2 out of my main 3 are f/1.4). Which of course brings up another point beyond the testable ISO question: full-frame and a 24 f/1.4 will add another 2 stops of low-light noise reduction over a 1.33x crop and an 18 f/3.8, simply by letting one use ISO 640 instead of 2500. Link to post Share on other sites More sharing options...
colorflow Posted October 21, 2009 Share #37 Posted October 21, 2009 Hans, really refreshing to see quantitative work like this on the forum. Really great work. If I may ask a few questions out of curiosity: I am not sure if your model separated out the various contributions to noise, i.e. RN, QN, Photon Shot Noise, Pixel Response Non Uniformity (PNRU) Noise and the Compression Noise, or were they lumped together? If yes, was QN a big contributor? Can that be lowered by deeper wells in the silicone? What is photon shot noise? Thanks Link to post Share on other sites More sharing options...
t024484 Posted October 21, 2009 Author Share #38 Posted October 21, 2009 Hans, really refreshing to see quantitative work like this on the forum. Really great work. If I may ask a few questions out of curiosity: I am not sure if your model separated out the various contributions to noise, i.e. RN, QN, Photon Shot Noise, Pixel Response Non Uniformity (PNRU) Noise and the Compression Noise, or were they lumped together? If yes, was QN a big contributor? Can that be lowered by deeper wells in the silicone? What is photon shot noise? Thanks all noise types are treated individually, since they all have completely different characteristics for different exposure and ISO. Read noise is indendent from exposure, but becomes more dominant on higher ISO. QN plays a minor role and is not dependent on exposure and not on ISO. Shot Noise is dependent on Exposure and on ISO. Compression noise is dependent on exposure but not on ISO. Etc, etc. For a better insight in photon shot noise, please look here: Shot noise - Wikipedia, the free encyclopedia Hans Link to post Share on other sites More sharing options...
colorflow Posted October 21, 2009 Share #39 Posted October 21, 2009 Thanks Hans. Link to post Share on other sites More sharing options...
patrick parker Posted October 21, 2009 Share #40 Posted October 21, 2009 There is one thing i don't understand about the one stop difference. Since it is the same sensor as M8 but larger, would it be true to say Leica is not releasing a new firmware for the M8 to maintain this difference? Link to post Share on other sites More sharing options...
Recommended Posts
Archived
This topic is now archived and is closed to further replies.