thebarnman Posted June 4, 2012 Share #281 Posted June 4, 2012 Advertisement (gone after registration) I don't believe you are going to see any more printable detail from MM files over M9 files due to default NR in a modern raw processor. Is there a way to turn that off OR what raw processors allow for use without NR? Link to post Share on other sites More sharing options...
Advertisement Posted June 4, 2012 Posted June 4, 2012 Hi thebarnman, Take a look here Survey: Your opinion about the new LEICA M MONOCHROM. I'm sure you'll find what you were looking for!
thebarnman Posted June 4, 2012 Share #282 Posted June 4, 2012 You don't know what we will get in the M10, but if you want sharper cleaner images then the answer as to be "yes" Not when compared to a MM since the M10 will have the color filers over the sensor like the M9 regardless of the resolution. Link to post Share on other sites More sharing options...
Jamie Roberts Posted June 4, 2012 Share #283 Posted June 4, 2012 Is there a way to turn that off OR what raw processors allow for use without NR? I'm not sure what you're asking, but yes--all of the professional raw processors I know let you turn off noise reduction. My point was something else entirely though. But carry on Sounds like you will have lots of fun here! Link to post Share on other sites More sharing options...
thebarnman Posted June 4, 2012 Share #284 Posted June 4, 2012 I'm not sure what you're asking, but yes--all of the professional raw processors I know let you turn off noise reduction. My point was something else entirely though. But carry on Sounds like you will have lots of fun here! Yes thanks, that's what I wanted to know. Link to post Share on other sites More sharing options...
mwilliamsphotography Posted June 5, 2012 Share #285 Posted June 5, 2012 The presets of SilverFX were *designed* to be used with RGB files. Till the MM, there was no monochrome digital file to work with Except B&W film scans ... or flatbed scans of silver-prints . The presets also work at luminance levels ... the color filters are there to work on color images being converted to B&W. What we don't know yet is the spectral response of a dedicated B&W sensor compared to a sensor with color interpolation. It seems the MM files aren't simply producing a desaturated look as evidenced by Jonos side-by-side M9 and MM test shot. The MM file looks to have a wider tonal spread that helps separate the response to the colors verses needing to adjust the color filters in post. I was reading the Leica blurb about the MM, and the histogram looks to be different than that of the M9, according to them it is more accurate and reliable of a judge ... so blown highlights would be a user issue, not the camera's. -Marc Link to post Share on other sites More sharing options...
Jamie Roberts Posted June 5, 2012 Share #286 Posted June 5, 2012 Except B&W film scans ... or flatbed scans of silver-prints . The presets also work at luminance levels ... the color filters are there to work on color images being converted to B&W. What we don't know yet is the spectral response of a dedicated B&W sensor compared to a sensor with color interpolation. It seems the MM files aren't simply producing a desaturated look as evidenced by Jonos side-by-side M9 and MM test shot. The MM file looks to have a wider tonal spread that helps separate the response to the colors verses needing to adjust the color filters in post. I was reading the Leica blurb about the MM, and the histogram looks to be different than that of the M9, according to them it is more accurate and reliable of a judge ... so blown highlights would be a user issue, not the camera's. -Marc Marc--you're right: the SilverFX film conversion presets are the ones I was talking about. They are clearly meant to be used with colour information. Silverfast would probably be the better program there, no? Yes, the program has luminance and contrast moves--and tinting--if it didn't, then there would be no point to including it at all But if you go into the details of the monochrome film conversions they all contain colour tweaking. Anyway, I really don't think it's a stretch to say that SilverFX wasn't designed for the monochrome film scanner market, let alone the silver print duplication market, do you? I think Silverfast would be the better program there, honestly. And honestly, you lost me a bit with "wider tonal spread..." The MM certainly has more DR at higher ISOs and consequent exposure latitude (the noise floor is lower so there's more shadow information as ISO increases), which you get at the expense of colour information. Along with increased detail, that's all I've seen in the DNGs. Whether that's enough, or printable, or film-like, or a good ROI, are other questions Anyway, from David Farkas's latest test, it seems that the MM is more green sensitive than the M9. I say "seems" because it's really hard to tell from the web test, in part because he tried to make the M9 look like the MM, instead of the other way 'round ) So I have no idea what kind of false colour was induced at comparison time on the M9 side. But the leaves in the plants all look much brighter in the MM shots, given what I'd assume is the same exposure, so it's the magenta / green response that's different. Have you seen the posts of the monochromatic response of the MM vs TMax? Link to post Share on other sites More sharing options...
AlanG Posted June 5, 2012 Share #287 Posted June 5, 2012 Advertisement (gone after registration) What we don't know yet is the spectral response of a dedicated B&W sensor compared to a sensor with color interpolation. -Marc Since the underlying sensor is the same, I'm trying to see why one could not come up with a standard conversion method that would make the M9 files match the MM files in all but resolution and noise. How could the removal of the Bayer filter increase the spectral response in any significant way in the visible spectrum? (The data Leica released shows it does not.) Do people really think the MM is opening some new door to b/w photography? The only thing I can think of is that the M9's filters might cost it a little bit of dynamic range due to the need to boost the signal a bit more than on the MM. (I don't buy into the Bayer filters only cutting out one stop.) This should be pretty easy to measure. Even if there is ultimately a tiny difference in spectral response capability in real life, I can't imagine this would be significant in actual use. And people are being very very picky about stuff like this without shooting their own tests to see if this really would matter to them. Link to post Share on other sites More sharing options...
jaapv Posted June 5, 2012 Share #288 Posted June 5, 2012 Dave Farkas' blog has some interesting remarks on the subject. To start with the three colors have different densities, and there are of course more green pixels than others,leading to different boosting at the pre-demosaicing stage. Link to post Share on other sites More sharing options...
AlanG Posted June 5, 2012 Share #289 Posted June 5, 2012 Dave Farkas' blog has some interesting remarks on the subject. To start with the three colors have different densities, and there are of course more green pixels than others,leading to different boosting at the pre-demosaicing stage. Yes, I figured as much. There is no free lunch. But since they can be mapped to pretty much anything for a nice color image, the same would apply to b/w images other than a bit more noise limiting the dynamic range a little (perhaps). Once you put a yellow, green, or red filter on the MM you are losing 2 - 3 stops bringing down the effective ISO to 40-80. So there are trade-offs for sure. However if a boosted signal is necessary to get the M9's ISO to 160 then I don't see why Leica didn't include a less boosted ISO of 80 in the M9 if it offered any improvement. (So it probably didn't or they had some technical reason why they couldn't do it.) I think almost all typical low light b/w film photography was done without filters and the tones fell where they may. The same for a lot of photojournalism. Whereas filters were commonly used for landscapes, architecture, and various commercial applications. A test of the M9 at 160 ISO and converting to b/w vs the MM with a green or red filter and the ISO set around 500-640 should prove interesting as it should drive the sensors about the same and you'd see the difference 18MP of b/w resolution alone makes. Link to post Share on other sites More sharing options...
Jamie Roberts Posted June 5, 2012 Share #290 Posted June 5, 2012 Dave Farkas' blog has some interesting remarks on the subject. To start with the three colors have different densities, and there are of course more green pixels than others,leading to different boosting at the pre-demosaicing stage. If I understand it correctly (and I'm sure someone who understands this better than I do can chime in) there is no "boosting" at the demosaic stage. From the raw Bayer interpretation, the colour is analysed and inferred. The blue and red pixels are interpolated (guessed) from their nearest neighbours, and since there's less of them than green ones in the standard pattern there is more interpretation in the algorithm for those colours. So colours can be (more or less) accurate, but they're not boosted in the sense of sensor gain or amplification. Another way to get colour, of course, without interpolation would be not to use a Bayer pattern at all but have three differently substrate filters (like the Foveon chip) or, indeed, sensors one each for red blue and green (as the old cinema digitals used to do). So for the colour sensor used in the M9, the native ISO is 160, because according to Leica the Bayer filter causes light loss of one stop overall. Again, if I understand Leica's claims, the native ISO of the MM is consequently 320 due to the lack of the same filter. Of course, as Alan says, you're going to lose something as you add filters on the lens end Link to post Share on other sites More sharing options...
mjh Posted June 5, 2012 Share #291 Posted June 5, 2012 Taking everything into account – IR filter, RGB filter densities, spectral response of the chip itself –, the green-sensitive sensor pixels are generally more sensitive than the red- or blue-sensitive ones. To compensate for this effect there is a boost for red and blue, although it doesn’t need to be applied at the demosaicing stage – most likely it will be applied before. Since the green-sensitive pixels are the most sensitive, their sensitivity determines the sensitivity of the sensor. In other words, the green-sensitive pixels of the M9 have a base sensitivity of ISO 160 while the sensitivity of the red and blue pixels is lower. When the monochrome sensor has a base sensitivity of ISO 320, this implies that a monochrome pixel has a sensitivity advantage of one f-stop compared to a green-sensitive pixel of the colour sensor. Link to post Share on other sites More sharing options...
brianv Posted June 5, 2012 Share #292 Posted June 5, 2012 They could put clear glass over the sensor, do an M9vnir, and have a base sensitivity closer to ISO500... Okay, I'll settle for monochrome... I wonder how much a company would charge to switch out the IR filter for a clear filter. Link to post Share on other sites More sharing options...
AlanG Posted June 5, 2012 Share #293 Posted June 5, 2012 Since the green-sensitive pixels are the most sensitive, their sensitivity determines the sensitivity of the sensor. In other words, the green-sensitive pixels of the M9 have a base sensitivity of ISO 160 while the sensitivity of the red and blue pixels is lower. When the monochrome sensor has a base sensitivity of ISO 320, this implies that a monochrome pixel has a sensitivity advantage of one f-stop compared to a green-sensitive pixel of the colour sensor. I got all that and I am not technically knowledgeable about the kinds of dyes used to make Bayer filters. The most simple way of putting it, I have a hard time understanding how removing red, green, and blue filters only gains one stop but placing red, green, or blue filters over the same monochrome sensor will lose 2-3 stops. But can a green color separation filter in the Bayer array, or red or blue filter, really be made that is so pure that it only loses one stop? And even if that were true, I can't see how that ends up being only a 50% total light loss unless these filters are not doing that good of a job separating the colors. Just as a starting point assuming perfect filters, if our theoretical "white" light is made up of equal amounts of red, green, and blue light, removing all but the green will cut out 2/3rds. The same thing for the pixels covered by red and blue filters. To picture this, imagine three colored flashlights all illuminating the same spot. Assuming they have equal brightness, when you remove two of them, don't you lose 2/3rds of the light? So before you even get into the filter having some neutral density, there is a difference between only having 33% of the light vs. needing 50% for it to be a one stop loss. So, the best I can imagine for this to work, if the sensor itself is more sensitive to green light, they can't boost the red and blue pixels in the MM because it doesn't have any way to differentiate those. So they must be cutting down the amount of green light that hits the sensor via a magenta filter... thus lowering the ISO that way and producing a more suitable spectral response. Link to post Share on other sites More sharing options...
mjh Posted June 5, 2012 Share #294 Posted June 5, 2012 The Bayer filter array isn’t dividing the spectrum into three neatly separated bands; rather the transmission curves of the filters overlap (as do the spectral response curves of the the three different cone cell types in the human eye). The green filter has the broadest spectral response curve which is the main reason why green pixels are more sensitive than the red and blue ones. Link to post Share on other sites More sharing options...
AlanG Posted June 5, 2012 Share #295 Posted June 5, 2012 The Bayer filter array isn’t dividing the spectrum into three neatly separated bands; rather the transmission curves of the filters overlap (as do the spectral response curves of the the three different cone cell types in the human eye). The green filter has the broadest spectral response curve which is the main reason why green pixels are more sensitive than the red and blue ones. Yes but how is this different from typical red, green and blue over the lens filters or other kinds of separation filters and how well can various colors be represented if there is too much overlap? It seems to me that the more they overlap, the less the saturation. Link to post Share on other sites More sharing options...
pico Posted June 5, 2012 Share #296 Posted June 5, 2012 They could put clear glass over the sensor, do an M9vnir, and have a base sensitivity closer to ISO500... Okay, I'll settle for monochrome... I wonder how much a company would charge to switch out the IR filter for a clear filter. Removing the IR filter from a camera that has a Bayer pattern layer will just make the camera almost useless for ordinary photography because the firmware will still try to interpret neighboring pixels, and if you do use the near-infrared spectrum you must be prepared for some pretty ugly outcomes, especially if a bodily unattractive woman in a sun dress walks into the scene on a bright sunny day. Google these words: IR photography x-ray Link to post Share on other sites More sharing options...
mjh Posted June 5, 2012 Share #297 Posted June 5, 2012 Yes but how is this different from typical red, green and blue over the lens filters or other kinds of separation filters and how well can various colors be represented if there is too much overlap? It seems to me that the more they overlap, the less the saturation. Without any overlap we could differentiate between exactly three hues – red, green, and blue. Only because the transmission curves do overlap we are able to finely distinguish thousands of hues. Saturation could possibly suffer if a single wavelength would incite a response in not just two but all three types of pixels (or cone cells, for that matter). But not necessarily so: wavelengths between about 450 and 500 nm (blue to cyan) are registered by all the cone cell types in the human eye, for example. Link to post Share on other sites More sharing options...
AlanG Posted June 6, 2012 Share #298 Posted June 6, 2012 I know there has to be overlap. But how can it function if the spectrum overlap is greater than with common color separation filters that eat up 2 to 3 stops? All I'm getting at is that I'd like to see an example of color sep filters that only cut out one stop. Link to post Share on other sites More sharing options...
mjh Posted June 6, 2012 Share #299 Posted June 6, 2012 All I'm getting at is that I'd like to see an example of color sep filters that only cut out one stop. This is all about the green filters and green-sensitive pixels detecting about one half of the available light sounds plausible to me. It is different for the red and blue filters. In any case, that’s how it is. The M Monochrom exists to prove it. Link to post Share on other sites More sharing options...
mwilliamsphotography Posted June 8, 2012 Share #300 Posted June 8, 2012 Another camera talked to death before it's even out. -Marc Link to post Share on other sites More sharing options...
Recommended Posts
Archived
This topic is now archived and is closed to further replies.