 |
 | 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 / October 2006The myth of the "smaller" 4/3rds lens
Ron wrote: > I've heard rumors that Olympus will begin releasing smaller and lighter > lenses within the next year. That is only by making lens surfaces more curved (to increase magnification or reduce it more) in a given space. Also, extensive and expensive use of ED and other exotic glasses helps allow this. But people have to remember that lenses cannot "shrink" beyond a certain point and it has NOTHING to do with sensor size. A 200mm f2 lens STILL needs a 100mm wide front element to claim it is 200mm f2 (200/100 = 2). The REASON it seems that Olympus can make "smaller" lenses is because a 200mm lens on a Full frame or 1.3-1.6 cropped sensor provides a wider angle of view than the 2x Olympus that in-turn allows Olympus to put more pixels into a given area of a scene, and resolve more detail for a given lens focal length (provided both the Olympus and the other brand have the same pixel count). So the proverbial 300mm lens on the 2x sensor functions like a 400mm lens on a FF camera. The reciprocal being that you cannot get as wide a view with an Olympus sensor as with a larger sensor because it would require making lenses with shorter focal lengths. The shortest lens I've ever seen was Nikon's 6mm which on a FF SLR has twice the field of view as a 4/3rds camera. But, if someone is thinking Olympus can produce a 300mm f2.8 lens any smaller than Canon, they would be dead wrong. The lens STILL needs a 107mm of clear front aperture to meet it's speed claim. There is ongoing research into ways to focus light for photographic use that does not totally rely on glass/plastic as we know it particularly for use with very small sensors. The technology of imaging is evolving in ways that will not rely on traditional materials just as digital sensors have replaced film for most uses. Bigger is not better even at this point in the development of digital photography and will come to be seen as the liability it has always been when better technical solutions become available. > There is ongoing research into ways to focus light for photographic use that > does not totally rely on glass/plastic as we know it particularly for use [quoted text clipped - 5 lines] > photography and will come to be seen as the liability it has always been > when better technical solutions become available. The flaw in this logic is number of photons. A smaller camera with a smaller aperture lens collects fewer photons. The signal-to-noise ratio in digital camera images is directly related to the square root of the number of photons collected in each pixel. Smaller pixels with smaller photons collected also results in smaller dynamic range. http://www.clarkvision.com/imagedetail/does.pixel.size.matter Roger >> There is ongoing research into ways to focus light for photographic >> use that does not totally rely on glass/plastic as we know it [quoted text clipped - 16 lines] > > Roger Roger, I don't think anybody implied a smaller aperture here. Yes, you'll always need a large area to collect photons from (the "front element"), but the rest of the "lens" (length etc) is not fixed by physical considerations. Take a look at this: http://prola.aps.org/abstract/PRL/v85/i18/p3966_1 Cheers! >>> There is ongoing research into ways to focus light for photographic >>> use that does not totally rely on glass/plastic as we know it [quoted text clipped - 17 lines] > Take a look at this: > http://prola.aps.org/abstract/PRL/v85/i18/p3966_1 hmmm <scratching head> http://physicsweb.org/articles/news/9/4/12 >>>> There is ongoing research into ways to focus light for photographic use >>>> that does not totally rely on glass/plastic as we know it particularly [quoted text clipped - 19 lines] > hmmm <scratching head> > http://physicsweb.org/articles/news/9/4/12 If it could be scaled to the dimensions required for photographic lenses (the uses described seem to be on a very short distance scale) then it might yield a lens sharper than current technology, but photographic lenses are seldom diffraction-limited at large apertures anyway, so it would seem to be a nonstarter. > >>> The flaw in this logic is number of photons. A smaller > >>> camera with a smaller aperture lens collects fewer photons... [quoted text clipped - 15 lines] > seldom diffraction-limited at large apertures anyway, so it would seem to be > a nonstarter. The idea is that it's a nontraditional way to focus light. > > >>> The flaw in this logic is number of photons. A smaller > > >>> camera with a smaller aperture lens collects fewer photons... [quoted text clipped - 17 lines] > > The idea is that it's a nontraditional way to focus light. I don't think this "holy grail" of refraction has been used across the visible spectrum yet. > > The idea is that it's a nontraditional way to focus light. > > I don't think this "holy grail" of refraction has been used across the > visible spectrum yet. I was merely trying to point out that various "physical limits" aren't as hard as they would appear. Others are, of course. "negative refractive index material" means that the speed of light is higher than c within that material, while this is not strictly impossible, it can only be obtained by quantum effects. And those usually have a major statistical behavior, which supposedly imposes noise. -Michael > "negative refractive index material" means that the speed of light is > higher than c within that material, while this is not strictly > impossible, it can only be obtained by quantum effects. And those > usually have a major statistical behavior, which supposedly imposes noise. c>1 is, in fact, strictly impossible (for the group velocity of light, which is what is of interest here). What "negative refractive index" means in this context is a bit more technical, and in Pendry's paper he actually considers a material for which the dielectric constant and the magnetic permeability (which describe the response of the material to applied electric and magnetic fields, eg light) are negative. Their product is still positive, and so the speed of light is not different from free space. Of course, this is an approximation, in actual experimental realisations it'll be slower. And reasoning that quantum effects always induce noise is simply untrue: Consider solids, which only exist because of quantum mechanical effects. Or, to take a more obvious example, a Bose-Einstein condensate: a purely quantum effect in which all the particles behave in an identical way even at finite temperature. Quite the opposite of what you'd expect! > > "negative refractive index material" means that the speed of light is > > higher than c within that material, while this is not strictly [quoted text clipped - 3 lines] > c>1 is, in fact, strictly impossible (for the group velocity of light, > which is what is of interest here). What "negative refractive index" Hmm.... The beginning of my reply disappeared, somehow. It should have read something like: The refractive index n is usually defined as the ratio of the speed of light in vacuo and in a material; thus, what you say would refer to n<1 rather than n<0. At any rate, c>1 is, in fact,....... Apologies. > The refractive index n is usually defined as the ratio of the speed of > light in vacuo and in a material; thus, what you say would refer to n<1 > rather than n<0. At any rate, > c>1 is, in fact,....... I read an article about this some weeks ago. The effect (n < 1 aka v>c) is obtained by something like this: The experiment is set to measure the time when a photon is detected. But a photon is a wave-like event. A pure wave starts at t=-infinity and last until t=+infinity. So a detectable photon is a superposition of waves of different frequencies, the phases of which determine the position of the photon. Thus the photon has no clear start or end, but in "reality" it starts and stops slowly. In the material with n<1 the waves are partly suppressed and partly amplified (by means of a "pumping" energy source). By this the detectable position of the photon is shifted in the process and thus it can move (seemingly) with v>c resulting in n<1. I seem to remember that as a result of the amplification process some a huge lot of noise (respecting n) was introduced in the experiment. -Michael > I read an article about this some weeks ago. The effect (n < 1 aka v>c) > is obtained by something like this: Hi, Can you tell me where you read it, or, if you remember, where this was carried out (ie in which university/laboratory)? > The experiment is set to measure the time when a photon is detected. But > a photon is a wave-like event. A pure wave starts at t=-infinity and [quoted text clipped - 8 lines] > amplification process some a huge lot of noise (respecting n) was > introduced in the experiment. It sounds like you are describing manipulation of individual frequency components of a pulse so as to change its shape. But how can you suppress a particular frequency component at once (everywhere)? Consider than one frequency component is, as you say, of infinite spatial extend; so if they were really able to shut it off at once, it would hardly be surprising that they would get instantaneous transfers of information! I suppose that there are various approximations and loose language being used here, though. Anyway, I'll try to find some information on this experiment. > Can you tell me where you read it, or, if you remember, where this was > carried out (ie in which university/laboratory)? I'll try to look it up. It was a German optical science Internet news letter. -Michael > There is ongoing research into ways to focus light for photographic > use that does not totally rely on glass/plastic as we know it [quoted text clipped - 5 lines] > photography and will come to be seen as the liability it has always > been when better technical solutions become available. You can't simply "get smaller" without reaching a point that bumps up against limitations rooted in the properties of light itself. Others here have discussed this at length (Think Roger and David Littleboy).  SignatureImages (Plus Snaps & Grabs) by Mark² at: www.pbase.com/markuson > You can't simply "get smaller" without reaching a point that bumps up > against limitations rooted in the properties of light itself. Others here > have discussed this at length (Think Roger and David Littleboy). Mark, When it comes to lenses, the only physical limitation is the relationship of the area from which light is collected ("front element") to the magnification of the image on the sensor. The rest is only determined by what we use to make the lenses (currently, glass). > There is ongoing research into ways to focus light for photographic use > that does not totally rely on glass/plastic as we know it particularly for [quoted text clipped - 5 lines] > photography and will come to be seen as the liability it has always been > when better technical solutions become available. While it is in principle possible to make a flat camera that from a magnification viewpont has a very long lens or one that from a depth of field and sharpness viewpoint has a very wide aperture using techniques developed for long-baseline interferometry, the difficulty with these techniques is that you don't get the sensitivity normally associated with those large apertures. I can't see this approach as being other than a niche solution. > There is ongoing research into ways to focus light for photographic use that > does not totally rely on glass/plastic as we know it particularly for use > with very small sensors. So when will we get these portable gravitic lenses? And will we need a huge truck for the power requirements, or will there be a hip-pocket fusion plant to go with it? > Bigger is not better even at this point in the development of digital > photography and will come to be seen as the liability it has always been > when better technical solutions become available. That must be why the APS films were such a rage and outsold everything else. Oh, no, wait, that's the reason every self-respecting photographer uses a Minox instead of small, medium or large format, and why all the digital cameras are modelled after the Minox cameras ... instead of these large rectangles, which you can hardly put into a small pocket or up your sleeve. -Wolfgang > Ron wrote: > > I've heard rumors that Olympus will begin releasing smaller and lighter [quoted text clipped - 5 lines] > people have to remember that lenses cannot "shrink" beyond a certain > point and it has NOTHING to do with sensor size. Why would you make such a point of this.. and then go about telling us the opposite? Of course sensor size makes a difference, *for a given angle of view*! And if you don't hold the angle of view constant for comparing whether a given camera/lens combination is bulkier than another, then what is the point?? Yes, we know about sensor noise issues, but this was about lens size/bulk. > A 200mm f2 lens STILL > needs a 100mm wide front element Umm. Forgive my pickiness - I'm a bit rusty on lens design. But is it the *front element* that gives the aperture calculation???? I mean I realise it has to be a certain size to allow the lens to operate to it's full 'f' capability, but isn't it the APERTURE that is used for calculating the f-ratio? Ie the IRIS opening, not the front element. I humbly apologise if I'm wrong, and I realise they are of course related, but let's be *accurate* here... > The REASON it seems that Olympus can make "smaller" lenses is > because a 200mm lens on a Full frame or 1.3-1.6 cropped sensor > provides a wider angle of view than the 2x Olympus that in-turn allows > Olympus to put more pixels into a given area of a scene, and resolve > more detail for a given lens focal length As per my initial comment - DUUUH! > So the proverbial > 300mm lens on the 2x sensor functions like a 400mm lens on a FF > camera. Am I missing something here as well? It's a 2x multiplication factor, and yet Rich says a 300 goes to ..400? Umm, yeah right... > The shortest lens I've ever > seen was Nikon's 6mm Sigh. Never looked at a video camera or a compact? > which on a FF SLR has twice the field of view as > a 4/3rds camera. Err, say what? I missed the point of that line.. Yes, to match that 6mm lens, Oly would have to create a 3mm. Challenging? Yes indeed. But how many 6mm (35 equiv) owners are there around here? (That 6mm lens, if mounted on the Oly, would act like a 12mm (35 equiv) extreme w/a, and I doubt *many* folks would be agonising that they couldn't go wider... > But, if someone is thinking Olympus can produce a 300mm f2.8 lens any > smaller than Canon, they would be dead wrong. But they *can* create a 150mm f2.8 that is smaller and is the EQUIVALENT lens in terms of field of view. Sigh. > The lens STILL needs a > 107mm of clear front aperture to meet it's speed claim. Now it's 'front aperture'? See comment above. Happy to be corrected... > > Ron wrote: > > > I've heard rumors that Olympus will begin releasing smaller and lighter [quoted text clipped - 22 lines] > I humbly apologise if I'm wrong, and I realise they are of course > related, but let's be *accurate* here... The general rule in lens design is you never use more aperture than you have to. Which means the front lens diameter (what isn't cut off by any retaining rings) is the aperture. However, it is possible that some lens designs have been made using a larger size than true aperture once internal body "stop downs" have been used. This would be used in a case where CA or some other aberration could be controlled in no other way, but it's rarely used because the cost of lens fabrication and materials rises exponentially with element diameter. The iris opening does dictate the aperture, but the iris in most lenses should open nearly as or as wide as the element's front aperture. Waste not.... > > The REASON it seems that Olympus can make "smaller" lenses is > > because a 200mm lens on a Full frame or 1.3-1.6 cropped sensor [quoted text clipped - 10 lines] > Am I missing something here as well? It's a 2x multiplication factor, > and yet Rich says a 300 goes to ..400? Umm, yeah right... No, the angle of view is 1/2 of that in a FF camera, so the 300mm on an FF = 600mm on the 2x. > > The shortest lens I've ever > > seen was Nikon's 6mm [quoted text clipped - 9 lines] > w/a, and I doubt *many* folks would be agonising that they couldn't go > wider... You'd be amazed how many FF Canon shooters regularly use 17mm or wider lenses. I'd venture very few 4/3rds owners will have a 8-9mm lens available. > > But, if someone is thinking Olympus can produce a 300mm f2.8 lens any > > smaller than Canon, they would be dead wrong. > But they *can* create a 150mm f2.8 that is smaller and is the > EQUIVALENT lens in terms of field of view. Sigh. That is true, but it is NOT the same as making a 300mm smaller than another 300mm. I was talking about lenses of the same f.l., not lenses that were equivalent in action on the two different sensors. > > The lens STILL needs a > > 107mm of clear front aperture to meet it's speed claim. > > Now it's 'front aperture'? See comment above. Happy to be corrected... As I said, in most lenses the front aperture IS the full aperture and determines the focal ratio. However, we are talking about normal and tele lenses and NOT wide angle lenses that have completely different characteristics. Which is why Olympus's old 8mm fisheye has a front element about 120mm wide. If you compare the old OM 8mm with the new one, the new one (made to support 4/3rds instead of FF) is substantially smaller. > The general rule in lens design is you never use more aperture than you > have to. > Which means the front lens diameter (what isn't cut off by any > retaining rings) is the aperture. Absolutely NOT. Take a look at a wide angle lens, and measure its front element, whilst examining the size of the actual *aperture* that you can see.... You were using the front element size to calculate/infer f-ratio. You *cannot* do that. > The iris opening > does dictate the aperture, but the iris in most lenses should open > nearly as or as wide as the element's front aperture. (Note how he has dropped to *most* lenses now...) Like I said, take a look through a wide angle.... > > > So the proverbial > > > 300mm lens on the 2x sensor functions like a 400mm lens on a FF [quoted text clipped - 5 lines] > No, the angle of view is 1/2 of that in a FF camera, so the 300mm on an > FF = 600mm on the 2x. Yes, Rich, in other words, YOUR calculation of 300x2=400 was wrong. It wasn't *mine*! Is it completely impossible for your eyes to see your mistakes? Can you not say "I was wrong"? > > > The shortest lens I've ever > > > seen was Nikon's 6mm [quoted text clipped - 11 lines] > You'd be amazed how many FF Canon shooters regularly use 17mm or wider > lenses. But *you* were referring to a 6. So how many FF Canon users shoot *that*? Your argument shifts like the wind - true trolling behaviour. > I'd venture very few 4/3rds owners will have a 8-9mm lens available. A minute ago it was a 3mm, but like I said... you'll change anything to try to keep the upper hand.. You are right, but I'm *not* arguing the fact that 4/3 currently has a restricted lens supply. Largely the 4/3 system has not been unsuccessful simply because the sensors haven't been good enough = not enough sales = not enough funds to develop a decent lens range. But *if* they manage/d to create a killer sensor, guess what would happen to the sales and then the lens situation. Frankly I think time is rapidly running out for 4/3. I'd like to jump to it, but they just aren't producing good high-res, low noise sensors. (I again point out that I would be intrigued to see what Fuji could do if given the chance..) > > > But, if someone is thinking Olympus can produce a 300mm f2.8 lens any > > > smaller than Canon, they would be dead wrong. [quoted text clipped - 4 lines] > another 300mm. I was talking about lenses of the same f.l., not lenses > that were equivalent in action on the two different sensors. But lenses of the same focal length are NOT the same in any meaningful way, as soon as you move to different sensor size!! Just answer the simple question - is the field of view of a 300mm on a FF the same as on a 4/3? No, of course it bloody well isn't. So if someone is looking for 300mm equivalence on a 4/3 system, s/he will only need 150mm for the same sort of photos. Tell me, Rich, is a 50mm a wide angle? Oh, I use MF by the way. So it is for me. Is a 21mm a wide angle? On my video camera that is quite zoomed in... Oh dear how confusing... (O; For the *equivalent usage*, the focal length measurement means NOTHING unless you specify the sensor size. You CAN make an equivalent lens smaller on a smaller sensored camera. End of story. It's not a myth. Yes, there are many other factors - if you want to talk sensor noise, then let's also talk about quality of glass, design of lens, processing engine, amplifier quality, brand of sensor, ad infinitum. Oh, by the way, I liked this addon: > As I said, in most lenses the front aperture IS the full aperture and > determines the focal ratio. However, we are talking about normal and > tele lenses and NOT wide angle lenses that have completely different > characteristics. GRIN. When exactly did 'we' start only talking normal and tele??? Oh, right, just *then* when you realised the error in your logic. > Which is why Olympus's old 8mm fisheye (This would be one of the 8mm lenses you referred to that no 4/3 users have, of course..) > has a front > element about 120mm wide. If you compare the old OM 8mm with the new > one, > the new one (made to support 4/3rds instead of FF) is substantially > smaller. Yes. Smaller. And you now say the front element cannot possibly be used to measure the aperture. Well, you've certainly proved your point there!!!??? Oh wait, no, you proved MINE. Thanks. What were those words again? Yes, here they are: > the new one (made to support 4/3rds instead of FF) is substantially > smaller. (O: PS - You still haven't changed THIS page to show it is a 5D image, NOT a 30D. Nor have you explained *why* you stripped off the EXIF data. One has to wonder if it was to prevent detection of the fraud... http://www.pbase.com/andersonrm/image/57631694 Hint, this ISN'T the *other* image you re-labelled, in case you have got yourself even more confused. More on this thread: http://groups.google.com.au/group/rec.photo.digital.slr-systems/browse_frm/threa d/408935ffbb912ccc/ Whoops, a small correction - (yes, I can happily admit *my* errors!): The line that says: "Largely the 4/3 system has not been unsuccessful" should (as will be clear from the context) have read: "Largely the 4/3 system has not been successful" Oh, how I abhor not having an absence of no double-negatives... (O: >>Ron wrote: >> [quoted text clipped - 24 lines] > I humbly apologise if I'm wrong, and I realise they are of course > related, but let's be *accurate* here... The accurate answer is the entrance pupil determines the true aperture. The front element must be at least the true aperture, or you don't get the full amount of photons. Typically, the front element is larger than the entrance pupil. The iris diaphragm may be much smaller as the lenses in front of the iris usually focus the light into a smaller cone. The entrance pupil is not necessarily a physical aperture. It gets quite complex in the many element lens designs for modern cameras. >>But, if someone is thinking Olympus can produce a 300mm f2.8 lens any >>smaller than Canon, they would be dead wrong. > > But they *can* create a 150mm f2.8 that is smaller and is the > EQUIVALENT lens in terms of field of view. Sigh. The problem with reducing the lens aperture is that it collects fewer photons. So in the equivalent field of view same megapixel count camera scaling, the 300 f/2.8 collects 4 times the number of photons as the 150 mm f/2.8. So the two cameras may produce the same resolution images, but the smaller camera image is 2x noisier for the same exposure time. Roger You *do* seem to be fixated on photon noise. (O; In my response, I said: >Yes, we know about sensor noise >issues, but this was about lens size/bulk. I thought that was sufficient, given the topic. Roger: > So the two cameras may produce the same resolution images, > but the smaller camera image is 2x noisier for the same exposure > time. But hang on - first up, we are talking about two completely different camera formats, and noise may or may not be an issue. Is noise an issue on a Hassy back? On a Canon 1DS MkII? On a Nikon D2X? On a P&S? Yes, of course at some point it becomes an issue, but you can't just blithely say it is 2x noisier and seemingly dismiss the concept. Given possible improvements in sensor design (- I seem to recall that when pushed you admit that there is probably at least another factor of 30% or so left for improvements in processing, more efficient sensor packing, different designs of wells, microlenses etc), and the fact that already many sensors *somehow* perform well beyond their weight (eg the Fuji sensor designs, the F30, the larger canon CMOSes) - I don't believe that 4/3 is too small to ever have a really good low-noise sensor, which is the implication.. In other words, it may indeed be "2x noisier", but if the noise is so low that a 2-fold increase is still very low-noise, who cares? In theory you could shrink the best Canon FF CMOS to 4/3... and it would be 2x noisier... Do you think anyone would be complaining much about those noise levels? Like I said, this post was about the alleged "myth of smaller 4/3 lenses". There is no myth. You can make a smaller lens, given a smaller sensor. *At some point* sensor noise may/will become an issue, but to say that 4/3 is that point reminds me of some Arthur C Clarke quotes: Any sufficiently advanced technology is indistinguishable from magic. Every revolutionary idea seems to evoke three stages of reaction. They may be summed up by the phrases: (1) It's completely impossible. (2) It's possible, but it's not worth doing. (3) I said it was a good idea all along. If an elderly but distinguished scientist says that something is possible, he is almost certainly right; but if he says that it is impossible, he is very probably wrong. > You *do* seem to be fixated on photon noise. (O; > [quoted text clipped - 16 lines] > Yes, of course at some point it becomes an issue, but you can't just > blithely say it is 2x noisier and seemingly dismiss the concept. Take a full frame DSLR & the half size 4/3 with lenses of equal field of view and aperture so the same amount of light from the same scene is hitting different sized sensors for the same exposure time. At first the 4/3 setup will look better/brighter than the underexposed full frame (I think) but as you increase the exposure time it's pixels will fill up, spill over & blow out while the full sized sensor keeps gathering information & producing lower noise results. PS the 4/3 setup will have more DOF at the same aperture & field of view and actually is more sensitive in low light (without adjusting ISO). The full frame will have the option of catching up with ISO as long as there aren't strong highlights to blow out. > Given possible improvements in sensor design (- I seem to recall that > when pushed you admit that there is probably at least another factor of [quoted text clipped - 27 lines] > possible, he is almost certainly right; but if he says that it is > impossible, he is very probably wrong. SignaturePaul Furman http://www.edgehill.net/1 Bay Natives http://www.baynatives.com >> You *do* seem to be fixated on photon noise. (O; >> [quoted text clipped - 29 lines] > full frame will have the option of catching up with ISO as long as there > aren't strong highlights to blow out. When you say "same aperture" are you talking about the same f/ratio or about same physical aperture, i.e. both have a physical aperture of say 100 mm? If you are talking about the same f/ratio, then the problems with the above are: The same amount of light from the same scene is _not_ hitting different sized sensors for the same exposure time. The same amount per unit area is hitting the sensor for each but not the same total amount. For the same resolution four times as many pixels have to be fitted into a unit area on the 4/3 camera so each gets a quarter as many photons. The 4/3 will _not_ "look better/brighter" nor will "its pixels fill up, spill over, and blow out". Instead for a given exposure the number of photons collected is not as far above the noise threshold. It will _not_ be "more sensitive in low light", it will be less so. Now, if you are talking about the same physical aperture, that means that the 4/3 has an f/ratio half that of the full frame. Now it gets as many photons on the frame as does the full frame, and four times as many per unit area, each pixel gets the same number as for the full frame (not _more_, just the same number). You will not use the same exposure on the 4/3 in that case, you will use 1/4 the exposure. This at first glance sounds good but to get the same number of photons per pixel as a Canon with the $75 50mm f/1.8 lens attached you'd need to use an f/0.9 25mm lens on the 4/3. While the smaller sensor size does somewhat simplify the len designer's problem, still, producing a f/0.9 lens with more that 3800 lines/mm of resolution is not a simple problem and that lens is not going to be cheap. To match the number of photons/pixel delivered by the f/1.2 Canon lenses your 4/3 lens would have to be an f/0.5 lens--this is getting way off into hyperexotic territory. >> Given possible improvements in sensor design (- I seem to recall that >> when pushed you admit that there is probably at least another factor of [quoted text clipped - 27 lines] >> possible, he is almost certainly right; but if he says that it is >> impossible, he is very probably wrong. >>Take a full frame DSLR & the half size 4/3 with lenses of equal field of >>view and aperture so the same amount of light from the same scene is [quoted text clipped - 18 lines] > sized sensors for the same exposure time. The same amount per unit area is > hitting the sensor for each but not the same total amount. Ah, gotcha. I've heard the claim that P&S do better in low light & thought I was on to something but probably mistaken. It does make sense to me that the physical aperture opening would have to be as large to gather as many photons so I think you are correct. > To match the number of photons/pixel delivered by > the f/1.2 Canon lenses your 4/3 lens would have to be an f/0.5 lens--this is > getting way off into hyperexotic territory. Yes, and I suppose that would be a comparable DOF too? If this was achieved, you would be into overflowing pixels but I agree it's impractical. P&S do often have pretty decent f/stop though like f/2.8 compared to f/4 on consumer zooms so there may be a bit of light gathering advantage there but P&S sensors are another level smaller than 4/3 so... SignaturePaul Furman http://www.edgehill.net/1 Bay Natives http://www.baynatives.com > You *do* seem to be fixated on photon noise. (O; > [quoted text clipped - 16 lines] > Yes, of course at some point it becomes an issue, but you can't just > blithely say it is 2x noisier and seemingly dismiss the concept. Well, it is 2x noisier. Whether or not that is important for a given application is irrelevant. But it is the fundamental reason why small pixel cameras have such poor high ISO performance. So, in the above example, the larger format camera has noise the same as the small format camera but at FOUR TIMES THE ISO! There are many applications where such differences are important. > Given possible improvements in sensor design (- I seem to recall that > when pushed you admit that there is probably at least another factor of [quoted text clipped - 4 lines] > don't believe that 4/3 is too small to ever have a really good > low-noise sensor, which is the implication.. Actually, I've stated about 3x. Current sensors are running about 30% quantum efficiency, and near 100% is possible (and it has been demonstrated with back side illuminated CCDs). But 3x still falls short of the 4x that the pixel size gives in the example being worked here. Higher QE sensors are much more expensive to produce, so are likely to appear in larger format cameras as opposed to cheap point and shoots, so the difference in ISO equivalency will show as an even larger factor (e.g. 3x in QE * 4x in area, so 12x higher ISO). > In other words, it may indeed be "2x noisier", but if the noise is so > low that a 2-fold increase is still very low-noise, who cares? In > theory you could shrink the best Canon FF CMOS to 4/3... and it would > be 2x noisier... Do you think anyone would be complaining much about > those noise levels? Noise is always an issue in low light/high speed photography as you are most likely photon noise limited. > Like I said, this post was about the alleged "myth of smaller 4/3 > lenses". There is no myth. You can make a smaller lens, given a [quoted text clipped - 3 lines] > > Any sufficiently advanced technology is indistinguishable from magic. The difference here is digital cameras have reached a fundamental limit of being photon noise limited. While very impressive, it is not magic. > Every revolutionary idea seems to evoke three stages of reaction. They > may be summed up by the phrases: (1) It's completely impossible. (2) [quoted text clipped - 4 lines] > possible, he is almost certainly right; but if he says that it is > impossible, he is very probably wrong. There are many counter examples that illustrates the above is absurd. And of course there are examples where it is correct too. In the case of photography, we are dependent on 1) sunlight or other natural light such as the moon and stars or 2) artificial light. All are finite, measurable and known. We can only hope that solar radiation does not increase enough to enable photographers to take shorter exposure pictures, as we will bake to death. ;-) Roger In article <452EFB12.50405@qwest.net>, "Roger N. Clark (change username to rnclark)" <username@qwest.net> wrote: > The problem with reducing the lens aperture is that it collects > fewer photons. So in the equivalent field of view same > megapixel count camera scaling, the 300 f/2.8 collects > 4 times the number of photons as the 150 mm f/2.8. What you are conveniently forgetting is that 24x36 is about 4 times the size of 4/3, which means you NEED 4 times the number of photons to even get the SAME exposure. a 4/3-sized part of the FF chip + 300mm will get exactly the same amount of light as the 4/3 camera with 150mm. (Now, there will be less, and larger, pixels on the 4/3 sized part of the FF-chip, but that is something different than the number of photons captured.) Next time when you have a presentation with a beamer, try moving it back so the projector makes an image twice as large. I bet it will be twice as dark too. Lourens > In article <452EFB12.50405@qwest.net>, > "Roger N. Clark (change username to rnclark)" <username@qwest.net> [quoted text clipped - 12 lines] > FF-chip, but that is something different than the number of photons > captured.) No, that is not correct. Total light gathered (total number of photons) is a function of aperture and only aperture per unit time. How many photons a pixel collects is a function of aperture and angular field of view per unit time. Given a 4/3 sensor versus a FF sensor with the same total number of pixels, so each camera has the same resolution, the FF pixels are proportionally larger. So given a 300 mm f/2.8 lens on FF, each pixel gathers 4x the photons of 150 mm f/2.8 lens on a sensor half the size. See: Digital Cameras: Does Pixel Size Matter? Factors in Choosing a Digital Camera http://www.clarkvision.com/imagedetail/does.pixel.size.matter and scroll down to f/ratio myth. Roger > Next time when you have a presentation with a beamer, try moving it back > so the projector makes an image twice as large. I bet it will be twice > as dark too. Not the same thing. In the above, you are concerned with photons per unit area, but in changing the camera size, you are changing the area of the pixel. Roger In article <45315C0B.3060805@qwest.net>, "Roger N. Clark (change username to rnclark)" <username@qwest.net> wrote: > No, that is not correct. Total light gathered (total number > of photons) is a function of aperture and only aperture per unit time. But aperture is relative to the fl. The "light gathered" is a function of the absolute aperture AND the angle of view. (focal length). It is quite obvious that a telephoto takes light from a smaller part of the subject than a wide lens, which is why the same (relative) aperture has to be a larger diameter on the telephoto. > So given a 300 mm f/2.8 lens on FF, each pixel > gathers 4x the photons of 150 mm f/2.8 lens on a sensor > half the size. Yes, because it is larger it needs more light. This could indeed influence signal-to-noise ratio to the advantage of larger sensors, but many other factors apply for example fill-factor and technology used. 4x5" was never any faster than 35mm, that should be a clue. (same sensor...) What I do know is that people who sell studio lighting, now sell a lot less of the really heavy 6400ws packs and such, because of digital. That is because sensors are smaller than 4x5 or 6x6 film used before. It needs LESS light. In similar fashion, 4/3 needs less light than FF. > Digital Cameras: Does Pixel Size Matter? > Factors in Choosing a Digital Camera > http://www.clarkvision.com/imagedetail/does.pixel.size.matter Don't find that page very impressive, sorry. You completely skip the fact that the image is projected behind the lens, and this projection is larger on FF, reducing the intensity of the image quadratically. Also the in-camera distances are larger, which actually costs light. Any 4x5" user knows that and it can be observed with any normal-focusing lens and a tripod. set to infinity, measure light, extend lens to minimum focusing distance. Exposure will be 1/3 or 1/2 stop less now. This stuff gets more complicated with internal-focusing lenses. (t-stop) > In the above, you are concerned with > photons per unit area, but in changing the camera size, > you are changing the area of the pixel. You don't get my point, I suppose. But I agree on that large-sensor camera's can have better signal-to-noise ratio, because of their size. It's just not as simplistic as many believe. In similar fashion, a smaller chip actually gives LESS DoF. But, the shorter fl and smaller diameter apertures (for the same image) give more DoF, and this effect is stronger. Lourens > You completely skip the fact that the image is projected behind the > lens, and this projection is larger on FF, reducing the intensity of the > image quadratically. I think this paragraph says all that needs to be said about your post. > In article <45315C0B.3060805@qwest.net>, > "Roger N. Clark (change username to rnclark)" <username@qwest.net> [quoted text clipped - 8 lines] > subject than a wide lens, which is why the same (relative) aperture has > to be a larger diameter on the telephoto. Aperture is NOT related to f/number. f/number is aperture divided by focal length. A 100 mm f/4 lens has an aperture diameter of 25mm, whereas a 200 mm f/4 lens has a 50mm aperture. The 50mm aperture collects FOUR time the light of the 25mm aperture. You can also have a 200 mm lens with a 25mm aperture. >>So given a 300 mm f/2.8 lens on FF, each pixel >>gathers 4x the photons of 150 mm f/2.8 lens on a sensor >>half the size. > > Yes, because it is larger it needs more light. It has nothing to do with "need," it is simply aperture gathers more light. It is like buckets in the rain: larger buckets collect more rain drops; the larger buckets do NOT NEED more rain drops. > This could indeed > influence signal-to-noise ratio to the advantage of larger sensors, but > many other factors apply for example fill-factor and technology used. > > 4x5" was never any faster than 35mm, that should be a clue. Never? Try taking pictures of stars. Here is a simple test. Go out at night and point your camera centered on the north star (northern hemisphere observers; southern hemisphere observers point to the south celestial pole). Put on your longest lens. Let's say its 300 mm f/5.6. Take a 30 second exposure. Now put on a short lens, like 50 mm and set it to f/5.6. record another 30 second exposure. Now examine which lens gave more stars. Hint: it is the larger aperture even though both were at the same f/ratio. Another hint: astronomers build large telescopes to collect more light, not to get more "telephoto" reach. Atmospheric turbulence limits resolution, especially on long exposures to about an arc-second or two. So, extending from the test above, the 3-meter aperture f/5.6 observatory telescope will record more stars with the same camera than the small lens test above, yet all work at f/5.6. The issue of "speed" and f/ratio is a balance of the amount of light collected by the lens and the focal length spreading that collected light over the focal plane. But how that light is collected in the focal plane is important too, and that relates to the size of the pixels. It is the variable size of pixels that changes the equation from film. > (same > sensor...) What I do know is that people who sell studio lighting, now > sell a lot less of the really heavy 6400ws packs and such, because of > digital. That is because sensors are smaller than 4x5 or 6x6 film used > before. It needs LESS light. In similar fashion, 4/3 needs less light > than FF. It is not NEED. It is capacity. Cameras don't "need" light. The pixels in small sensor cameras simply can't hold many electrons (converted photons), as larger pixels. So digital camera manufacturers have defined the exposure to give a signal using less photons, but at the expense of noise. >> Digital Cameras: Does Pixel Size Matter? >> Factors in Choosing a Digital Camera >> http://www.clarkvision.com/imagedetail/does.pixel.size.matter > > Don't find that page very impressive, sorry. Fine, you are welcome to your opinion. > You completely skip the fact that the image is projected behind the > lens, and this projection is larger on FF, reducing the intensity of the > image quadratically. No I don't. I consider three critical factors: aperture, focal length and pixel size. In film days, the film (sensor) did not change between formats. That is not true with digital. If you consider two cameras, each with the same megapixel count, so having the same spatial resolution, the larger format camera has larger pixels. Those larger pixels compensate for the magnification of the longer focal length of the scaled up camera. But since the aperture of the lens is larger, more photons are collected and each pixel in the larger format camera collects more photons. > Also the in-camera distances are larger, which actually costs light. Any > 4x5" user knows that and it can be observed with any normal-focusing > lens and a tripod. set to infinity, measure light, extend lens to > minimum focusing distance. Exposure will be 1/3 or 1/2 stop less now. > This stuff gets more complicated with internal-focusing lenses. (t-stop) What you describe with lens distance is changing magnification. Magnification spreads light out. That has nothing to do with format size between cameras, like APS-C, 1/1.8", FF. >>In the above, you are concerned with >>photons per unit area, but in changing the camera size, [quoted text clipped - 3 lines] > camera's can have better signal-to-noise ratio, because of their size. > It's just not as simplistic as many believe. And why you do you think a larger format camera has a higher signal-to-noise ratio? If you consider film cameras, say 35mm to 4x5, with the same film in both, signal to noise does not change. Resolution changes. With digital camera when the resolution is constant between formats (same field of view plus same megapixel count), pixel size changes, thus the equation changes. In a digital camera, scaling the format up but keeping resolution the same, S/N increases because aperture increases delivering more photons to each larger pixel. > In similar fashion, a smaller chip actually gives LESS DoF. But, the > shorter fl and smaller diameter apertures (for the same image) give more > DoF, and this effect is stronger. The above statement is confusing. A larger format camera gives a narrower depth of field at widest aperture (e.g. at f/4 on 35mm camera has a shallower depth of field than f/4 on a small chip P&S digital camera). The larger format camera has a larger range of depth of field possible. For maximum depth of field (small aperture), the small format digital camera has no advantage. The larger format camera can stop its aperture down to the same diameter as the small camera, and deliver the same depth of field with the same signal-to-noise ratio. Roger > ... > It has nothing to do with "need," it is simply aperture [quoted text clipped - 16 lines] > give a signal using less photons, but at the expense > of noise. P&S digital often has ISO 50 right? So they are more capable of capturing an image in low light (apparently) though yes the cost is more noise & less dynamic range. Probably you could underexpose a DSLR image and push it to ISO 50 with similar noise. >> You completely skip the fact that the image is projected behind the >> lens, and this projection is larger on FF, reducing the intensity of [quoted text clipped - 58 lines] > > Roger SignaturePaul Furman http://www.edgehill.net/1 Bay Natives http://www.baynatives.com >> ... >> It has nothing to do with "need," it is simply aperture [quoted text clipped - 21 lines] > less dynamic range. Probably you could underexpose a DSLR image and push > it to ISO 50 with similar noise. ??????? I think your understanding of ISO numbers is backward. The larger the ISO number the less light is needed for the exposure. ISO 50 is a setting for very bright light. ISO 1600 would be a "low light" setting. <snip> > > In article <45315C0B.3060805@qwest.net>, > > "Roger N. Clark (change username to rnclark)" <username@qwest.net> [quoted text clipped - 144 lines] > > Roger It's amazing that understanding the nuances of photography does nothing to educate people on even the basic rules of how lenses operate. But there is something I wonder about. If a small CCD has a "too small" well capacity, instead of making the pixels larger, what about making them deeper? They just tested a tiny Samsung with a 10 meg CCD on dpreview. I would LOVE it if they shot same images, 100 ISO against a Nikon D80, which also has 10 megs of very different CCD. > It's amazing that understanding the nuances of photography does > nothing > to educate people on even the basic rules of how lenses operate. Or sensors, or reading a website review. > They just tested a tiny Samsung with a 10 meg CCD on dpreview. I > would > LOVE it if they shot same images, 100 ISO against a Nikon D80, which > also has 10 megs of very different CCD. They did. Samsung NV10 - several test images using the same targets as the Nikon D80 (and every other camera they test) for resolution and noise. Your lack of observational skills is showing again... >>> In article <45315C0B.3060805@qwest.net>, >>> "Roger N. Clark (change username to rnclark)" <username@qwest.net> [quoted text clipped - 153 lines] > a "too small" well capacity, instead of making the pixels larger, > what about making them deeper? ??? A point on a CCD could be a MILE deep, but it will still only gather the light that falls on its face. SignatureImages (Plus Snaps & Grabs) by Mark² at: www.pbase.com/markuson |
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.