 |
 | Home | Contact Us | FAQ | Search & Site Map | Link to Us |
 | Sign In | Join | Other 45 Sites in Network |
| PhotoKB HomeDiscussion Groups Digital Photography Digital PhotoDSLR CamerasZLR CamerasPoint & Shoot CamerasFilm Photography 35 mmLarge FormatMedium formatDarkroomFilm and LabsOther EquipmentPhoto Technique Nature PhotographyPeople PhotographyTechnique GeneralGeneral Photo Topics General TopicsAustralian PhotographyUK PhotographyDirectoryPhoto Clubs |
Photo Forum / Digital Photography / DSLR Cameras / March 2006Canon's problem with WA lenses and FF (the solution?)
Lets set aside any question about the sensor and microlenses because in truth, they produce only minor issues with given lenses. It is the LENSES that are the problem. So, its clear that this is a problem, another poster confirmed it by posting a response from Canon indicating that WA lenses do NOT produce a large enough fully-illuminated circle to cover the FF sensors. A gross example would be if you tried to use a 35mm lens to illuminate a medium format sensor or film. You'd get a circle of illumination inside the outline of the sensor. The other aberrations (coma, etc) that plague the edges of these lenses have been addressed by people adopting superior lenses from Leica to replace their Canon WAs. Given the resolution of the FF Canon sensors, what choice did they have? But how could Canon fix this problem? Would it only require a re-design of their 35mm film lenses to increase field illumination or would they (in addition) have to scrap the old bayonette they use (with its small opening) and adopt one with a wider opening to accommodate new lens designs? Lastly, will this require something radical, like elimination of the reflex mirror to allow the rear lens elements to be closer to the sensor itself? Rich, since you don't own any Canon gear why do you care? I use my cameras on a daily basis now and a lot of film before that. Vignetting was never an issue then and it's not now because you fix it in the dark room or Photoshop. I'll be buying the following Canon Wide's this year and this so-called problem doesn't bother me in the least. Nor does it any photographer who is out taking pictures and not sitting at their computer practicing mental masturbation with a non-problem. 35mm f/1.4L, 24mm f/1.4L and 16-35mm f/2.8L. And if you really want to see edge fall off you should take a look at what the Peleng 8mm I ordered today will do on a 5D. ************************************************************** "There has always been war. War is raging throughout the world at the present moment. And there is little reason to believe that war will cease to exist in the future. As man has become increasingly civilized, his means of destroying his fellow man have become ever more efficient, cruel and devastating. Is it possible to put an end to a form of human behavior which has existed throughout history by means of photography? The proportions of that notion seem ridiculously out of balance. Yet, that very idea has motivated me. James Nachtwey War Photographer http://www.jamesnachtwey.com/ > Rich, since you don't own any Canon gear why do you care? He doesn't care. It should be obvious by now. He is nothing but a troll.  SignatureCharles > Rich, since you don't own any Canon gear why do you care? You don't own any olympus gear but bash it regularly. I guess only people talking about a canon problem should STFU? SignatureStacey >> Rich, since you don't own any Canon gear why do you care? > >You don't own any olympus gear but bash it regularly. I guess only people >talking about a canon problem should STFU? I just think the 4/3 sensor is a bad idea and try to prevent people from wasting money on it. No, you can talk about Leica and Leaf back problems if you wish. Signature"There has always been war. War is raging throughout the world at the present moment. And there is little reason to believe that war will cease to exist in the future. As man has become increasingly civilized, his means of destroying his fellow man have become ever more efficient, cruel and devastating. Is it possible to put an end to a form of human behavior which has existed throughout history by means of photography? The proportions of that notion seem ridiculously out of balance. Yet, that very idea has motivated me. James Nachtwey War Photographer http://www.jamesnachtwey.com/ >>> Rich, since you don't own any Canon gear why do you care? >> [quoted text clipped - 3 lines] > I just think the 4/3 sensor is a bad idea and try to prevent people > from wasting money on it. And he thinks canon's wide lenses are bad and people shouldn't waste their money on them, I tend to agree. SignatureStacey Don't own, but what if I ever do? Wasn't that your argument for not buying "digital" lenses, that you were going FF eventually? Nothing is carved in stone and therefore people should at least be marginally concerned about the state of products. > Lets set aside any question about the sensor and microlenses because in > truth, > they produce only minor issues with given lenses. It is the LENSES > that are the > problem. <bullshit snipped> Since you don't own a camera, why do you give a sh.t? > Lets set aside any question about the sensor and microlenses because in > truth, > they produce only minor issues with given lenses. It is the LENSES > that are the > problem. All lenses? You are really the master of the grand statement, Rich. The champion. However, to drop a not of reality into your fantasy, read this: http://www.luminous-landscape.com/reviews/lenses/24-70-review.shtml > So, its clear that this is a problem, another poster confirmed it by > posting a response > from Canon indicating that WA lenses do NOT produce a large enough > fully-illuminated > circle to cover the FF sensors. er, read the above again ... A gross example would be if you tried > to use a 35mm lens > to illuminate a medium format sensor or film. You'd get a circle of > illumination inside the > outline of the sensor. Oh dear, that means my plans to use my 21mm lens on my 8x10 view camera won't work? Thanks, Rich, you saved me the trouble of trying it! The other aberrations (coma, etc) that plague > the edges of these > lenses have been addressed by people adopting superior lenses from > Leica to replace > their Canon WAs. Given the resolution of the FF Canon sensors, what > choice did > they have? Generalities, generalities ... a few people might have tried Leica lenses - and lost all lens-camera communication. Are their results any better? they haven't been too noticeable, at least in these groups. > But how could Canon fix this problem? Would it only require a > re-design of their 35mm film > lenses to increase field illumination or would they (in addition) have > to scrap the old bayonette they use (with its small opening) and adopt > one with a wider opening to accommodate new lens designs? What??? The Canon EOS mount has the widest throat and the shortest register distance of any current 35mm or digital slr camera. Lastly, will > this require something radical, like elimination of the reflex mirror > to allow > the rear lens elements to be closer to the sensor itself? The closer the rear element is to the film the more acute the ray angles will be at the edges. Getting the lens closer isn't the problem. getting one Rich to admit he posts unadulterated bullshit is the *real* problem. Colin D. > Lets set aside any question about the sensor and microlenses because in > truth, > they produce only minor issues with given lenses. It is the LENSES > that are the > problem. Actually, most experienced photographers do not see it as a problem, and it is certainly not limited to Canon, and it has nothing to do with the fact that the camera is digital. You get vignetting with wide angle lenses on film, too. People forgot that when digital cameras with small sensors came out because all the existing lenses had image circles designed for 35mm film. When manufacturers designed lenses for the smaller sensors or they made the sensors bigger, guess what? The vignetting that film photographers had always know came back. Only with a bunch of inexperienced newbie whiners with no film history now using digital cameras, you got a lot of crybabies boo-hooing over this "design problem." Sure, you could make the image circle larger with any lens. But then photographers would complain that they are unable to use the whole image that the lens produces, even if that image is not optimum from edge to edge. Experienced photographers also know that vignetting gets worse if you stick a filter on the lens and know how to adjust for that. They don't do idiotic things like stick two filters on the camera or put a "protective" UV filter on a wide-angle lens unless they don't care about the vignetting. Pay attention to the guys who have been using cameras for more than three years. They might actually know what they are talking about. > Actually, most experienced photographers do not see it as a problem, > and it is certainly not limited to Canon, and it has nothing to do [quoted text clipped - 7 lines] > history now using digital cameras, you got a lot of crybabies > boo-hooing over this "design problem." Which came first, the chicken or the egg? I agree with what you are saying, but think about it a bit. Your argument and mine demonstrate the total uselessness of an FF sensor since this is nothing more than nomenclature for a dead format. It's totally irrelavent what the sensor size is just as long as long as both the lens system and body work properly together. To me, this means that I shouldn't be plagued with vignetting of the old "film" days. The "solution" for both Canon and Nikon is to design their lens system to work properly with whatever nomenclature they want to slap on their sensors. Most "experienced" photographers realize what lens characteristics they had to "live with" back in the day, but they shouldn't be expected to endure it in the 21st century. This statement holds true, especially when Canon and Nikon expect their customers to trash a $5,000+ dSLR body every 12-18 months and buy the latest and greatest cutting edge technology. To me, it's important to buy into a lens system that I can keep pretty much for life and throw bodies away every year. > Sure, you could make the image circle larger with any lens. But then > photographers would complain that they are unable to use the whole [quoted text clipped - 4 lines] > camera or put a "protective" UV filter on a wide-angle lens unless > they don't care about the vignetting. See above: They can keep the smaller sensor size and vignetting isn't too much of an issue. Of course, with the crop factor they would need to design a high quality WA to fill in the gap. I agree with you about the filters, but that is the photographer's choice. If a person wants to slap a filter on the lens they are more than welcome to do so, but that is no reason for a lens manufacturer to offer lenses that vignette on whatever "new" format they offer. > Pay attention to the guys who have been using cameras for more than > three years. They might actually know what they are talking about. True, but why continually administer CPR on a dead horse when it's time to send it to the butcher? Rita > Which came first, the chicken or the egg? I agree with what you are > saying, [quoted text clipped - 19 lines] > and > throw bodies away every year. ALL photographic lenses from ALL manufacturers (excluding some scientific and very expensive lenses) have light fall off. Some projection lenses have low, but non zero light fall off. In fact, a zero light fall-off lens apparently as not yet been designed, but it is scientifically possible. See: Lens Systems with Large Optical Invariant J. Brian Caldwell, SPIE, 5865-16 V. 2, 2005. A true zero fall-off lens including considering angle of incidence on the lenses with super AR coatings has yet to be done. The size of a wide angle lens of short focal length would be similar to very large telephoto lenses, and with a cost of many thousands of dollars, and very high weight. The FF goal is maximizing the image quality in many respects for the given format. Larger formats always produce a higher resolution image, all things being equal. The 35mm size format is a good compromise between quality, speed and portability. For other factors in image quality, and far more important, IMHO, than light fall off (which is correctable) is pixel size. High megapixel count for high resolution plus large pixels equal large sensor. That will not change. See: Digital Cameras: Does Pixel Size Matter? Factors in Choosing a Digital Camera http://www.clarkvision.com/imagedetail/does.pixel.size.matter Roger > ALL photographic lenses from ALL manufacturers (excluding > some scientific and very expensive lenses) have light fall off. > Some projection lenses have low, but non zero light fall off. > In fact, a zero light fall-off lens apparently as not yet been > designed, but it is scientifically possible. See: True, I never disputed this. Using the "sweet spot" by cropping the image or using a smaller sensor will reduce or eliminate most of the problem. > The FF goal is maximizing the image quality in many respects for > the given format. Larger formats always produce a higher resolution > image, all things being equal. The 35mm size format is a good > compromise between quality, speed and portability. Yes, and all of these add compromises, which may or may not be good for everyone. > For other factors in image quality, and far more important, > IMHO, than light fall off (which is correctable) is pixel size. [quoted text clipped - 4 lines] > Factors in Choosing a Digital Camera > http://www.clarkvision.com/imagedetail/does.pixel.size.matter Great read. Of course this all changes with advancements in technology. We have all seen it in all forms of the semiconductor market. A good example is basic CPUs. Look at the physical size and capabilities of the classic Pentium compared to the newest chips on the market. The same with sensors, I think they will advance to the point where density and performance will be greatly increased. Personally, I think this technology is just getting started and we will be seeing many improvements in the next decade. Rita >>See: >> [quoted text clipped - 3 lines] > >Great read. Of course this all changes with advancements in technology. No it doesn't. Physics is independent of technology, and physics is what makes small pixels worse than big pixels. SignatureKennedy Yes, Socrates himself is particularly missed; A lovely little thinker, but a bugger when he's pissed. Python Philosophers (replace 'nospam' with 'kennedym' when replying) >> Digital Cameras: Does Pixel Size Matter? >> Factors in Choosing a Digital Camera [quoted text clipped - 9 lines] > greatly increased. Personally, I think this technology is just getting > started and we will be seeing many improvements in the next decade. The problem with your assertion is we are already at physical limits: 1) There are finite numbers of photons, making pixels smaller means less photons per pixel. 2) Diffraction is a fundamental limit already limiting lens resolution, 3) quantum efficiencies are already in the >~ 30% percent range, which is one of the few areas that can be improved, but no more than 3x. 4) If QEs do improve, it is likely that both large and small sensors would be improved, maintaining the quality gap. Roger >> http://www.clarkvision.com/imagedetail/does.pixel.size.matter > [quoted text clipped - 4 lines] >I think they will advance to the point where density and performance will be >greatly increased. Only if someone invents smaller photons, or makes the Sun emit four or eight times as many photons per second. Of course the first being fundamental physics won't happen; the second is inconceivable, and lethal if it did happen. SignatureIan G8ILZ >Only if someone invents smaller photons, or makes the Sun emit four or >eight times as many photons per second. Of course the first being >fundamental physics won't happen; the second is inconceivable, and >lethal if it did happen. Actually, far from being inconceivable, the second is inevitable and certainly will be lethal but, hopefully, won't occur for a few million years yet. ;) SignatureKennedy Yes, Socrates himself is particularly missed; A lovely little thinker, but a bugger when he's pissed. Python Philosophers (replace 'nospam' with 'kennedym' when replying) >>Only if someone invents smaller photons, or makes the Sun emit four or >>eight times as many photons per second. Of course the first being [quoted text clipped - 4 lines] >certainly will be lethal but, hopefully, won't occur for a few million >years yet. ;) From my (rather hazy) recollections of astrophysics, the Sun is only about 70% of the Chandrasekhar limit above which a star goes nova. Below this, is it not destined to just shrink into a white dwarf. Or did you mean the helium flash, which might occur at some point? Not that we need to worry about it - unless we are looking for *very* long-term archival storage... David SignatureDavid Littlewood >>>Only if someone invents smaller photons, or makes the Sun emit four or >>>eight times as many photons per second. Of course the first being [quoted text clipped - 14 lines] > >David We should be very happy the sun is a small, mundane little star. Were it a blue giant (assuming we could live on a planet near a star like a blue giant) it would go supernova in as little as 20 million years. We wouldn't even be here! As it is we probably have a couple billion more good years left! -Rich >>>Only if someone invents smaller photons, or makes the Sun emit four >>>or eight times as many photons per second. Of course the first being [quoted text clipped - 7 lines] >From my (rather hazy) recollections of astrophysics, the Sun is only >about 70% of the Chandrasekhar limit above which a star goes nova. Yes, the Chandrasekhar limit is about 1.4 solar masses. >Below this, is it not destined to just shrink into a white dwarf. Or >did you mean the helium flash, which might occur at some point? I was thinking more of the red expansion phase - at which point the inverse square law will make the incident light on the earth's surface from the sun much greater than it currently is. ;-) Either way, it isn't going to happen soon. SignatureKennedy Yes, Socrates himself is particularly missed; A lovely little thinker, but a bugger when he's pissed. Python Philosophers (replace 'nospam' with 'kennedym' when replying) >>>>Only if someone invents smaller photons, or makes the Sun emit four >>>>or eight times as many photons per second. Of course the first being [quoted text clipped - 16 lines] >inverse square law will make the incident light on the earth's surface >from the sun much greater than it currently is. ;-) OK, yes, I see that happens first - the helium flash occurs when the star moves from red giant to white dwarf. >Either way, it isn't going to happen soon. Unlikely the human race will be around, for sure. David SignatureDavid Littlewood >>Only if someone invents smaller photons, or makes the Sun emit four or >>eight times as many photons per second. Of course the first being [quoted text clipped - 4 lines] >certainly will be lethal but, hopefully, won't occur for a few million >years yet. ;) Ah, well, y-e-s. I was thinking on the time scale that matters to posters here, not one that might not even mater to the Human species. SignatureIan G8ILZ >> Which came first, the chicken or the egg? I agree with what you are >> saying, [quoted text clipped - 28 lines] >Lens Systems with Large Optical Invariant >J. Brian Caldwell, SPIE, 5865-16 V. 2, 2005. We are not dealing here with a problem of lens physics concerning vignetting. That is not the issue! Its like blaming a lens for showing diffraction, you can't do anything about it. You are simply making excuses that DO NOT APPLY if the lenses had been designed NOT to vignette a given sensor size. Yes, all lenses vignette, but they do NOT all produce the same image cicle size and THAT is the problem, the Canon WAs do not produce a LARGE enough image circle for the FF sensor size. Its easy to fix, they just won't do it! IF the Canon mount aperture is part of the problem, they should FIX that too. But at the very least, given the COST of L-lenses they could at least FIX some of the aberrations those lenses demonstrate (that are NOT demonstrated, apparently by other lenses swapped onto the Canons) at the edge of the field when used wider than a specific aperture. -Rich >>ALL photographic lenses from ALL manufacturers (excluding >>some scientific and very expensive lenses) have light fall off. [quoted text clipped - 21 lines] > wider than a specific aperture. > -Rich I'm sorry, but this is not correct. It is just canon bashing and from other posts, you don't even own a canon DSLR, so how would you know? Did you even bother to look up the reference and see what is really required to produce such a lens? If you did you would see NO camera manufacturer produces the ideal lens you seem to bash canon for not having. And then, if you could actually buy one, you would find it wouldn't be easy to use as it would be similar in size to current 300 mm f/2.8 lenses. Roger >>>ALL photographic lenses from ALL manufacturers (excluding >>>some scientific and very expensive lenses) have light fall off. [quoted text clipped - 24 lines] >I'm sorry, but this is not correct. It is just canon bashing >and from other posts, you don't even own a canon DSLR, so how would you know? Reading posts (not on this forum) from Canon owners who solved the problem by using non-Canon lenses instead? >Did you even bother to look up the reference and see what is really >required to produce such a lens? If you did you would see NO >camera manufacturer produces the ideal lens you seem to bash >canon for not having. Ideal lens? No. But one that performs to a specific level above what we've seen from Canon. And then, if you could actually buy one, >you would find it wouldn't be easy to use as it would be similar >in size to current 300 mm f/2.8 lenses. > >Roger A 20mm WA or thereabouts would need to be the size of a 300mm f2.8 to provide a full circle of illumination? Nonsense. I'd be fascinated to know why you think this would be the case when there are WA medium format lenses that do exactly what I'm talking about (produce an illumination circle large enough to EASILY cover their format) that are no where near the size you claim a FF (35mm) lens would be that could achieve the same thing. -Rich >>>>ALL photographic lenses from ALL manufacturers (excluding >>>>some scientific and very expensive lenses) have light fall off. [quoted text clipped - 27 lines] > Reading posts (not on this forum) from Canon owners who solved > the problem by using non-Canon lenses instead? What lens does not have a large enough image circle? Are you confusing image circle with light fall off? If an image circle is not large enough, you get black corners with no image. Light fall off produces darker corners, but there is still light, so can be mathematically corrected. On wide angle lenses, often if you stack more than one filter, on the lens, the corners get blocked, reducing the image circle. But I know of no lenses whose image circle is inadequate to fill the frame (except by definition a fisheye). I have never seen one, but I have observed light fall off from every lens I have ever used from multiple manufacturers. >>Did you even bother to look up the reference and see what is really >>required to produce such a lens? If you did you would see NO [quoted text clipped - 19 lines] > that could achieve the same thing. > -Rich No, almost any simple lens design can provide a full circle of illumination on a 35mm format. But all currently purchasable lenses from all manufacturers have significant light fall off. The paper I referred to sowed designs to minimize light fall off, but the designs would be huge lenses. So, you must be more clear about what you are ranting about. Image circle, or light fall off. All camera lenses have light fall off. I know of no Canon wide angle lens with too small of an image circle, nor such a lens from other manufacturers. The EF-S lenses used on a full-frame sensor would be an example of a lens with too small of an image circle. The unspecified wide angle lenses you cite have light fall off, regardless of manufacturer. Roger > Which came first, the chicken or the egg? I agree with what you are > saying, > but think about it a bit. Your argument and mine demonstrate the total > uselessness of an FF sensor since this is nothing more than nomenclature > for > a dead format. Ahem. As pixel counts continue to mount, the size of the photosites becomes increasingly important for the control of noise and allowance of high ISOs. Were it otherwise, then there would be no interest in MF digital cameras which have sensors on the order of 48 x 36mm, which is HUGE in terms of manufacturing the chips to do so. When it comes to any sensor size, bigger is always better. Fast glass lovers have always known this and have always paid the price. Signature-- r.p.e.35mm user resource: http://www.aliasimages.com/rpe35mmur.htm -- r.p.d.slr-systems: http://www.aliasimages.com/rpdslrsysur.htm -- [SI] gallery & rulz: http://www.pbase.com/shootin -- e-meil: Remove FreeLunch. > Ahem. As pixel counts continue to mount, the size of the photosites > becomes increasingly important for the control of noise and allowance > of high ISOs. You assume, of course, that future sensors will be using today's technology, which will definitely not be the case. Improved technologies and manufacturing techniques will improve S/N for any given size sensor. > Were it otherwise, then there would be no interest in MF digital > cameras which have sensors on the order of 48 x 36mm, which is HUGE > in terms of manufacturing the chips to do so. No argument that bigger is always better to a point. Everyone definitely would want the biggest most densely packed sensor that is as noise free as possible in a dSLR sized body. Of course, there is always going to be a limitation on physical size and performance of the lenses offered if you want to push the envelope. > When it comes to any sensor size, bigger is always better. Fast glass > lovers have always known this and have always paid the price. It all depends on the tradeoffs of manufacturing techniques, technology, and economics. I agree a bigger sensor is better for eliminating the crop factor, but if the same can be done with a smaller and more densely packed sensor than I'm all for it. Rita >It all depends on the tradeoffs of manufacturing techniques, technology, and >economics. No it doesn't. Ultimately it depends on the discovery of a negative index of refraction glass which, although possible in theory, has never been manufactured in the visible waveband. You can get such devices at radio wavelengths - indeed four of my colleagues invented the very first such material in the microwave band about 10 years ago. But to date nothing has indicated such an approach would be possible in the visible band. Until then, diffraction limits small pixels. Charge storage space limits small pixels. Noise limits small pixels. Your task for this years Nobel Prize in Physics is... SignatureKennedy Yes, Socrates himself is particularly missed; A lovely little thinker, but a bugger when he's pissed. Python Philosophers (replace 'nospam' with 'kennedym' when replying) > Your task for this years Nobel Prize in Physics is... Splitting the beer atom. Seriously, if this is the case then we are at the physical limitations of what we can squeeze out of the sensor for a given size. Since you guys have more experience in this area than I do I will have to take your word for it. I guess I will just sit by the sidelines and see what, if anything, transpires in the way of improved sensors. Thanks, I learned a lot. Rita >> Your task for this years Nobel Prize in Physics is... > [quoted text clipped - 6 lines] > Thanks, I > learned a lot. I think this year's PMA show is a good indicator of the matured technology: nothing really new concerning major strides in cameras with new sensors was introduced, unlike the previous numerous years. Roger In message <441E4730.1010908@qwest.net>, "Roger N. Clark (change username to rnclark)" <username@qwest.net> wrote: >>> Your task for this years Nobel Prize in Physics is... >> [quoted text clipped - 6 lines] >> Thanks, I >> learned a lot. >I think this year's PMA show is a good indicator of the matured >technology: nothing really new concerning major strides in cameras >with new sensors was introduced, unlike the previous numerous years. What they need to do now is eliminate banding, readout noise at lower ISOs, and perhaps increase the bit depth, at least for the lowest ISO on the camera. The 1DmkII, 20/30D, and 5D could all benefit from clean 16-bit readout. What I find striking, looking at noise performance on my 20D, is that Canon can readout the lower 1/16 of the sensor's linear DR fairly cleanly at ISO 1600, except for banding. The banding itself, in a blackframe, has only a tiny fraction of the standard deviation of the random noise, and does not alter the total standard deviation more than 1%, when subtracted from the blackframe, but is still very visible. The same camera does a pathetic job at ISO 100, when using the same absolute 1/16 sensor DR. I would expect more posterization at ISO 100, but posterization alone can't account for the blotchy mess, as posterizing the RAW capture at ISO 1600 to the same level of ISO 100 for the same absolute signal only degrades the image moderately. This means that the camera creates a tremendous amount of readout noise, relative to absolute sensor signal, at ISO 100. The standard deviation of a 20D blackframe at ISO 100 is about 2.07, and ISO 1600 is about 4.7. This means that, not even considering the extra posterization, ISO 1600 has about 7x the signal-to-noise ratio as ISO 100 pushed to ISO 1600. It would be *very* interesting, if we could have 16-bit ISO 100 data from current sensors that looks like current ISO 1600 for the 12 least significant bits. Here is my latest dramatization of the issue. The colors are flat because they're still RAW (but white-balanced), and there is no sharpening at all (less than "0" in RAW converter terms). The ISO 1600 image encompasses 320 RAW levels (out of 3967), and the ISO 100 image, 20 levels. EI for both is approximately ISO 10,000: http://www.pbase.com/jps_photo/image/57149515/original Signature<>>< ><<> ><<> <>>< ><<> <>>< <>>< ><<> John P Sheehy <JPS@no.komm> ><<> <>>< <>>< ><<> <>>< ><<> ><<> <>>< > The same camera does a pathetic job at ISO 100, when using the same > absolute 1/16 sensor DR. I would expect more posterization at ISO 100, > but posterization alone can't account for the blotchy mess, as > posterizing the RAW capture at ISO 1600 to the same level of ISO 100 for > the same absolute signal only degrades the image moderately. Isn't this just due to the analog gain being applied at ISO 1600? Your examples look pretty much as I would expect, but maybe I'm missing what you're getting at. SignatureJeremy | jeremy@exit109.com >> The same camera does a pathetic job at ISO 100, when using the same >> absolute 1/16 sensor DR. I would expect more posterization at ISO 100, >> but posterization alone can't account for the blotchy mess, as >> posterizing the RAW capture at ISO 1600 to the same level of ISO 100 for >> the same absolute signal only degrades the image moderately. >Isn't this just due to the analog gain being applied at ISO 1600? >Your examples look pretty much as I would expect, but maybe I'm missing >what you're getting at. The sensor wells are fully capable of what the 1600 gives, for the least significant 12 bits in a hypothetical ISO 100 16-bit readout. The noise in the ISO 100 push is not just the result of posterization; posterizing the ISO 1600 image to 20 RAW levels like the ISO 100 image only degrades it slightly at 100%; If you turned away, looked back, and it was posterized in the interim; you would probably not notice. That means that the bulk of the noise in the deepest ISO 100 shadows of the ISO 100 push is due to analog readout noise, which is then heightened by posterization. It is *NOT* coming from the sensor wells; it is readout noise. With better readout, current sensor wells could give results like the ISO 1600 from ISO 100. Signature<>>< ><<> ><<> <>>< ><<> <>>< <>>< ><<> John P Sheehy <JPS@no.komm> ><<> <>>< <>>< ><<> <>>< ><<> ><<> <>>< >>>The same camera does a pathetic job at ISO 100, when using the same >>>absolute 1/16 sensor DR. I would expect more posterization at ISO 100, [quoted text clipped - 20 lines] > better readout, current sensor wells could give results like the ISO > 1600 from ISO 100. What is the distinction between readout noise vs sensor well noise? > The sensor wells are fully capable of what the 1600 gives, for the least > significant 12 bits in a hypothetical ISO 100 16-bit readout. The noise [quoted text clipped - 6 lines] > the ISO 100 push is due to analog readout noise, which is then > heightened by posterization. Yes, I see what you're saying. I've long suspected that if the digitization was actually able to use everything the sensor could provide, that there would be no real point to increased ISO settings at all -- ISO 800 would be nothing more than chopping off part of the ISO 100 data, assuming a "native" sensitivity of 100 (which probably wouldn't be the case, but they would likely have to normalize it to the closest traditional rating). But I was thinking the limit was just in the digitization. What's the readout noise all about, and why would it be different for ISO 100 than 1600? SignatureJeremy | jeremy@exit109.com >What's the readout noise all about, and why would it be different for ISO >100 than 1600? The standard deviation of the noise in an ISO 1600 frame is only 2.27x what it is in an ISO 100 frame on the 20D. That noise exists relative to the fraction of the sensor's range used at that ISO (typically about 100% at ISO 100, and 6.25% or 1/16 at ISO 1600). So, if you calculate this noise relative to the sensor range that ISO 1600 covers, you have to divide the 2.27 by 16, for ISO 100. 2.27 / 16 = 0.14; that means that the *absolute* signal to noise ratio at ISO 100 is 0.14 what it is at ISO 1600. That extra noise has to come from somewhere, and it isn't there in the ISO 1600 images, so either some really good NR is done to the RAW data at ISO 1600, or the shadows of the ISO 100 have noise that emerges from reading the sensor, and not a necessary part of the sensor signal. The RAW files are written way to fast to have any kind of real noise reduction, and the noise looks genuine enough (random and rich at the nyquist) at ISO 1600. When I get a little time, I am going to repeat this scenario with more options. I will compare like I already have, plus show a posterized version of the ISO 1600, and a very clean ISO 100 "exposed to the right, and the same posterized like the ISO 100 deep shadows. Signature<>>< ><<> ><<> <>>< ><<> <>>< <>>< ><<> John P Sheehy <JPS@no.komm> ><<> <>>< <>>< ><<> <>>< ><<> ><<> <>>< >In message <441E4730.1010908@qwest.net>, >"Roger N. Clark (change username to rnclark)" <username@qwest.net> [quoted text clipped - 49 lines] > >http://www.pbase.com/jps_photo/image/57149515/original I think some of it (not all) depends on the lighting of the image. But there are definitely radical differences between the way cameras handle this kind of thing. -Rich http://www.pbase.com/andersonrm/olympus_e1_tests >I think some of it (not all) depends on the lighting of the image. >But there are definitely radical differences between the way >cameras handle this kind of thing. Uh, no ... we're counting photons here (or trying to). If the lighting is the same for both images, then the deepest shadows only differ by digitization method. Signature<>>< ><<> ><<> <>>< ><<> <>>< <>>< ><<> John P Sheehy <JPS@no.komm> ><<> <>>< <>>< ><<> <>>< ><<> ><<> <>>< [] > Here is my latest dramatization of the issue. The colors are flat > because they're still RAW (but white-balanced), and there is no [quoted text clipped - 3 lines] > > http://www.pbase.com/jps_photo/image/57149515/original Fascinating, John! More quantisation noise in the ISO 100 image, and poorer sharpness because (presumably) of the wider aperture! Thanks for sharing. David Results depend on the DSLR/sensor. What I've seen on the E-1 (smaller sensor, CCD not CMOS) is higher grain levels on the pushed 1600 ISO image but a flatter contrast on the pushed 100 ISO image. Both pushed to 10000. http://www.pbase.com/andersonrm/olympus_e1_tests In message <LFPTf.39473$wl.12091@text.news.blueyonder.co.uk>, "David J Taylor" <david-taylor@blueyonder.co.not-this-bit.nor-this-part.uk> wrote: >Fascinating, John! More quantisation noise in the ISO 100 image, and >poorer sharpness because (presumably) of the wider aperture! Same aperture, same shutter speed. Only ISO setting varied (exposure index was ISO 10,000 in both cases). The problem is not just quantization. I posterized the 1600 shot to the same number of RAW levels as the ISO 100 image, and it deteriorated a bit, but still looked more like the ISO 1600 image. No blotchy streaks like the ISO 100; just random noise and slight thin banding. Signature<>>< ><<> ><<> <>>< ><<> <>>< <>>< ><<> John P Sheehy <JPS@no.komm> ><<> <>>< <>>< ><<> <>>< ><<> ><<> <>>< >In message <LFPTf.39473$wl.12091@text.news.blueyonder.co.uk>, >"David J Taylor" [quoted text clipped - 7 lines] > >The problem is not just quantization. Exactly, it is read out noise too, hence the systematic structure. SignatureKennedy Yes, Socrates himself is particularly missed; A lovely little thinker, but a bugger when he's pissed. Python Philosophers (replace 'nospam' with 'kennedym' when replying) > Which came first, the chicken or the egg? I agree with what you are > saying, > but think about it a bit. Your argument and mine demonstrate the total > uselessness of an FF sensor since this is nothing more than nomenclature > for > a dead format. Since the majority of lenses on the market today for use on SLR digital cameras were designed for 35mm sized sensors (film), it is hardly "useless." You're German, aren't you, Rita? I've noticed a tendency for Germans to speak in absolutes when speaking or writing in English. ("Useless," "foolish." No grey area.) My wife's mother and grandmother did just that, as did her grandfather on her mother's side. That leads to a confrotational tone to almost every discussion in which there is a disagreement. >It's totally irrelavent what the sensor size is just as long > as long as both the lens system and body work properly together. To me, [quoted text clipped - 15 lines] > and > throw bodies away every year. Well, Nikon has shown a reluctance/inability to do just that when designing DX lenses, if the reports and reviews I read have any credence. And Canon's EF-S lenses have a tendency to vignette, too. I wonder what it is about lens designers and their designs that won't allow them to keep that to a minimum. >> Sure, you could make the image circle larger with any lens. But then >> photographers would complain that they are unable to use the whole [quoted text clipped - 14 lines] > lens manufacturer to offer lenses that vignette on whatever "new" format > they offer. It only seems to be no issue when the smaller sensor is used in conjunction with lenses designed for larger sensors. When the designers are presented with an opportunity to design for the smaller sensor, they seem driven to produce the same results that the old designs had with film and 35mm sized sensors. Even Oly, their 35-100, for instance, is actually larger than the Canon 70-200 f2.8, for which it is the equivalent. That's how they seem to have avoided light fall off, build a lens larger than it needs to be... >> Pay attention to the guys who have been using cameras for more than >> three years. They might actually know what they are talking about. > > True, but why continually administer CPR on a dead horse when it's time to > send it to the butcher? 'Cause, m'dear, the horse is not only alive, but kicking... SignatureSkip Middleton http://www.shadowcatcherimagery.com > > Actually, most experienced photographers do not see it as a problem, > > and it is certainly not limited to Canon, and it has nothing to do [quoted text clipped - 15 lines] > this means that I shouldn't be plagued with vignetting of the old "film" > days. <snip> That's not an entirely true statement. Sensor size, apart from sensel size and s/n, is important to depth of field considerations, and the 35mm FF format with the existing supply of lenses is pretty well ideal for being able to use differential or shallow focus, and yet afford adequate dof when needed for deep scenes. APS-C sensors are barely adequate for differential focus, and smaller sensors still just have too much dof to be professionally useful. Vignetting was always with us, but was less important with optical printing, because the enlarging process also showed lens vignetting which at least partly cancelled the negative vignetting, so everybody was happy. A point about software corrections for the 'problem' - if it is really a problem at all. I use reasonably high-end software that among other things corrects for vignetting. The sky in a print, for example, is dead even across the print. But, it looks wrong, it looks too light at the corners. I have to burn in the corners slightly to make the image hold together, to keep the eye in the picture. My conclusion is that the technical and complete elimination of vignetting may suit the techies, but it's not good for the image producers who want a pleasing and harmonious picture. There's moderation in all things, including vignetting. Colin D. >> Lets set aside any question about the sensor and microlenses because in >> truth, [quoted text clipped - 13 lines] >digital cameras, you got a lot of crybabies boo-hooing over this >"design problem." But you know what is being left out? The fact those horrible WA Canon images (that Canon posted when the 5D appeared!) suffered from worse edge problems than vignetting, problems that can't be corrected in software. -Rich |
Reply to this Message |
 Sign In
 Join
 My Latest Posts My Monitored Threads My Blog My Photo Gallery My Profile My Homepage Start New Thread Enable EMail Alerts Rate this Thread
|
©2009 Advenet LLC Privacy Policy - Terms of Use
This website includes both content owned or controlled by Advenet as well as content owned or controlled by third parties.