Prev: SI Facescape
Next: FF camera with mirrorless design
From: Alfred Molon on 29 Apr 2010 09:16 In article <290420100216076231%nospam(a)nospam.invalid>, nospam says... > > Obviously Bayer is not more accurate. And how could it be, with just 1/3 > > of the needed information? > > obviously, based on what? Perhaps because Bayer only captures 1/3 of the necessary information? > delta-e measurements show foveon having *less* accurate colour than > bayer. foveon also has a lot more noise than bayer, particularly at > higher isos. Hint: full-colour is not equal to Foveon. I'm not commenting on the technical performance of the Foveon implementation. > > A full colour sensor measures luminance at the pixel level. Bayer does > > not. > > foveon does not because it doesn't measure rgb. *both* require > processing. A full-colour sensor *does* measure all three colour components. Free your mind and stop thinking of Foveon. There are also other full- colour implementations: negative scanners, cameras with three CCDs, that Hassy medium format camera which used pixel-shift techniques to capture all three colour components. > maybe a mythical future sensor might do better than either foveon or > bayer, but that doesn't currently exist. Other implementations exist, see examples above. > > > > In any case we were talking about *spatial* interpolation, which Bayer > > > > does to generate the final image. > > > > > > nope. it fills in the holes. > > > > With guesses, which often are wrong. > > it's not guessing, it's calculated, and it's quite accurate. see above, > regarding delta-e measurements. You cannot calculate what numbers will win the lottery next week. -- Alfred Molon ------------------------------ Olympus E-series DSLRs and micro 4/3 forum at http://tech.groups.yahoo.com/group/MyOlympus/ http://myolympus.org/ photo sharing site
From: nospam on 29 Apr 2010 11:23 In article <MPG.2643a4ff741f109a98c2bd(a)news.supernews.com>, Alfred Molon <alfred_molon(a)yahoo.com> wrote: > > > Obviously Bayer is not more accurate. And how could it be, with just 1/3 > > > of the needed information? > > > > obviously, based on what? > > Perhaps because Bayer only captures 1/3 of the necessary information? yet it manages to produce a more accurate result. jpeg images have even *less* 'necessary information' and are often indistinguishable from the original. > > delta-e measurements show foveon having *less* accurate colour than > > bayer. foveon also has a lot more noise than bayer, particularly at > > higher isos. > > Hint: full-colour is not equal to Foveon. I'm not commenting on the > technical performance of the Foveon implementation. foveon is the only existing full colour sensor. maybe one day something better will come along but as of *right now* bayer provides the best quality. > > > A full colour sensor measures luminance at the pixel level. Bayer does > > > not. > > > > foveon does not because it doesn't measure rgb. *both* require > > processing. > > A full-colour sensor *does* measure all three colour components. with that definition, foveon is not a full colour sensor. > Free your mind and stop thinking of Foveon. There are also other full- > colour implementations: negative scanners, cameras with three CCDs, that > Hassy medium format camera which used pixel-shift techniques to capture > all three colour components. all of which have serious drawbacks. for example, shifting the sensor means you can't take photos of anything that moves, i.e., just about all real world photography. > > maybe a mythical future sensor might do better than either foveon or > > bayer, but that doesn't currently exist. > > Other implementations exist, see examples above. not for cameras. > > > With guesses, which often are wrong. > > > > it's not guessing, it's calculated, and it's quite accurate. see above, > > regarding delta-e measurements. > > You cannot calculate what numbers will win the lottery next week. that has absolutely nothing whatsoever to do with a camera sensor or interpolation.
From: Ray Fischer on 29 Apr 2010 12:59 Alfred Molon <alfred_molon(a)yahoo.com> wrote: >In article <290420100216076231%nospam(a)nospam.invalid>, nospam says... > >> > Obviously Bayer is not more accurate. And how could it be, with just 1/3 >> > of the needed information? >> >> obviously, based on what? > >Perhaps because Bayer only captures 1/3 of the necessary information? Which isn't true, of course. There is a lot of redundant information in images and a lot of what the Bayer sensor omits is unneeded. And please don't argue about that. Simply noting the effectivesness of jpeg compression should be enough to show that images contain a lot of redundant information. -- Ray Fischer rfischer(a)sonic.net
From: Alfred Molon on 29 Apr 2010 13:05 In article <290420101123540833%nospam(a)nospam.invalid>, nospam says... > > Perhaps because Bayer only captures 1/3 of the necessary information? > > yet it manages to produce a more accurate result. Perhaps it does not. > jpeg images have even *less* 'necessary information' and are often > indistinguishable from the original. jpeg is a data compression method. You are mixing up things. > foveon is the only existing full colour sensor. So? Full colour cameras exist, which is what matters. > maybe one day something better will come along but as of *right now* > bayer provides the best quality. No. For instance this Hassy with sensor shift manages to capture the full colour information and outperforms Bayer cameras. > all of which have serious drawbacks. for example, shifting the sensor > means you can't take photos of anything that moves, i.e., just about > all real world photography. Buildings for instance don't move, but the point is that full colour cameras outperform Bayer cameras. > > > maybe a mythical future sensor might do better than either foveon or > > > bayer, but that doesn't currently exist. > > > > Other implementations exist, see examples above. > > not for cameras. Yes, for instance the sensor shift Hassy. > > > > With guesses, which often are wrong. > > > > > > it's not guessing, it's calculated, and it's quite accurate. see above, > > > regarding delta-e measurements. > > > > You cannot calculate what numbers will win the lottery next week. > > that has absolutely nothing whatsoever to do with a camera sensor or > interpolation. It does, because that is exactly what Bayer does: interpolate missing values, by assuming that they are a function of the data in the neighbouring pixel. These are guesses, which often are wrong. You cannot calculate the value of a random variable. -- Alfred Molon ------------------------------ Olympus E-series DSLRs and micro 4/3 forum at http://tech.groups.yahoo.com/group/MyOlympus/ http://myolympus.org/ photo sharing site
From: nospam on 29 Apr 2010 13:41
In article <MPG.2643dae57b79e95198c2be(a)news.supernews.com>, Alfred Molon <alfred_molon(a)yahoo.com> wrote: > > > Perhaps because Bayer only captures 1/3 of the necessary information? > > > > yet it manages to produce a more accurate result. > > Perhaps it does not. there is no perhaps about it. the accuracy can be measured. delta-e (colour errors) for foveon is higher, or less accurate. or just look at the images. they're full of weird colour casts, including yellow skin, cyan skies, etc. and it's not consistent even among multiple foveon cameras. since there's a huge variability, it *can't* be more accurate. > > jpeg images have even *less* 'necessary information' and are often > > indistinguishable from the original. > > jpeg is a data compression method. You are mixing up things. so is bayer. 1/3 the data for an almost perfect reproduction. > > foveon is the only existing full colour sensor. > > So? Full colour cameras exist, which is what matters. but not a full colour sensor. > > maybe one day something better will come along but as of *right now* > > bayer provides the best quality. > > No. For instance this Hassy with sensor shift manages to capture the > full colour information and outperforms Bayer cameras. only if the subject isn't moving, and where are the tests that show it's better? > > all of which have serious drawbacks. for example, shifting the sensor > > means you can't take photos of anything that moves, i.e., just about > > all real world photography. > > Buildings for instance don't move, but the point is that full colour > cameras outperform Bayer cameras. buildings may not move but people and animals do, which is what most people photograph. even taking a handheld photo of a non-moving building is going to be harder since the camera has to take multiple exposures. once it does all that, you then have 4x the data to move, which means it's going to be slower and use up more storage. many drawbacks for very little gain. > > > > > With guesses, which often are wrong. > > > > > > > > it's not guessing, it's calculated, and it's quite accurate. see above, > > > > regarding delta-e measurements. > > > > > > You cannot calculate what numbers will win the lottery next week. > > > > that has absolutely nothing whatsoever to do with a camera sensor or > > interpolation. > > It does, because that is exactly what Bayer does: interpolate missing > values, by assuming that they are a function of the data in the > neighbouring pixel. These are guesses, interpolation is *not* guessing. > which often are wrong. what is the error rate? exactly how often is it wrong? > You cannot calculate the value of a random variable. it's not random. |