 |
 | Home | Contact Us | FAQ | Search & Site Map | Link to Us |
 | Sign In | Join | Other 45 Sites in Network |
Natural Science Forum / Physics / Optics / September 2004Lens with fixed focal
I look for a 35mm SLR Nikon mount lenses for a Fuji S2 with a very particular requeriment, focal lenght must be constant when focusing at least in the 3 meters - infinty range, for arquitectural photography (photogrammetry). Secondary requeriments are price, auto-focus, luminosity, overall quality, and so as usual, I apreciate them, but the key factor is "fixed focal lenght". A zoom lens could also fit if used only in the wide and tele end positions if the focal lenght does not changes when focusing. For example, Sigma 5.6 28-200 IF Lens is a good lens, but at 200 position the real focal lenght varies from around 165 mm focusing at 10 meters to 190mm focusing near infinity, you can see how magnification varies when focusing. This maybe because its internal focus system, good for others but a big problem for me. I purchased this lens and gave me lots of problems because this until I noticed and now I almost can't use it. On the other side a very cheap Nikon 1.8 50mm lens is my best one for this prupouse. The focual lenght varies quite little (around 0.2mm from the real 52.7mm) and image quality and luminosity is very good. Also works fine Cosina 3.5 19-35mm when used in 19mm position. This one has some distortion but that is no problem while the focal lenght doesnt vary too much with focus range. I was told to look for fixed focal, manual focus, and extension focus (Cosina 19 and Nikon 50 are extension focus lenses, I mean, they move frontal lens when focusing) but im not sure what lens to order. Im thinking in purchasing some vivitar cheap lenses just for trying as there is not much money and could worth the while, but Im afraid of purchasing quality lenses when then might be useless for me. Price is not a problem, but I dont want to spend 2000 Euro in a lens that becomes worst (for this) than a 100 Euro one... Any recomendation? Any website or catalog where I can check this caracteristics or any clue o keyword to look for. Sorry, but you really want to go to one of the photography newsgroups for this question. We deal mostly with the basics of optics, not the application for photographic cameras of commercially available lenses. -- Bob May Losing weight is easy! If you ever want to lose weight, eat and drink less. Works every time it is tried! >Sorry, but you really want to go to one of the photography newsgroups for >this question. We deal mostly with the basics of optics, not the >application for photographic cameras of commercially available lenses. Well, some of us do have some interest in both. To be fair to the original poster, he is asking a question which is outside the field of interest for most photography enthusiasts. Using standard serial production equipment is often a good way to get precision optical components at a good price. The answer to the original question is that this is a specification which is normally not given explicitly in the manufacturer's technical documentation. Most, but not all, fixed focus length camera lenses focus by moving the entire optical assembly. The two main exceptions are internal focusing (IF) lenses (mostly expensive long focal length lenses) and lenses with close range correction by moving groups of elements (mostly wide angle retrofocus lenses such as the Nikkor 2.8/24mm). IMHO, the best way to ensure that the lens focuses entirely by moving the optical assembly is to use something like a bellows unit for focusing. and a suitable lens, possibly one without a focusing mount (such as the long discontinued 100mm Bellows-Nikkor) As a side issue I'm somewhat intrigued by the assertion that the Nikkor 1.8/50mm changes focal length by 0.2mm over its focusing range. Where does that number come from and is that figure really larger than the measurement error?  Signature- Helge Nareid Nordmann i utlendighet, Aberdeen, Scotland For e-mail, please use my initials with the same domain as the posted address > Well, some of us do have some interest in both. I agree that the question was made in a way that could fit better at rec.photo, but let me reformulate the problem to fit this grupo better and ask a few basic cuestions by the way. My apologies if this still don't fit here, i will not insist if there is no interest. Forget about particular models and makers, I will need to find it comercially but my problem first is to understand what do I need exactly and how to ask for it, and need help even in the definition of the problem. I have formation in physics, and work in development of photogrammetric software and tecniques for close range aplications (that is, not aerial). In this field real photos are modeled as point proyections from world into CCD, with a model for compensating radial and decentering distortion. The ideal lens for this is a pinhole, but is too impractical, you need too much light and exposure time. I need a real lens that resembles a pinhole at long distances so focal length does not change. Distortion is not a problem, can be calibrated and corrected. The optics should be as fixed as possible, nothing should move between taking two pictures. In practice I can model some tolerance for variations in focal length, distortion, and position of the optic axis between photos compared with calibration, but the final results are better when this parameters change the little. Also there is a problem with calibration charts. I can take images from a know 30 cm card placed at 1 m distance and get precise calibration values for this parameters. But when the model is 30 meters long at 100 m distance (ie, a building) distortion and optic axis positions doesn't move too much (and have little impact in final solution) but focal uses to change a lot and this is important. > To be fair to the original poster, he is asking a question which is > outside the field of interest for most photography enthusiasts. Using > standard serial production equipment is often a good way to get > precision optical components at a good price. Anyway I asked at rec.photo.digital and checked forums that optiker recomends, but I think I it is not something what architectural photographers or fuji customers uses to know, but some guidelines related to lens design and/or/for photogrammetry. I'm not even sure what to ask for exactly, remember, but is something of quite technical nature. > The answer to the original question is that this is a specification > which is normally not given explicitly in the manufacturer's technical > documentation. Most, but not all, fixed focus length camera lenses True. > focus by moving the entire optical assembly. The two main exceptions > are internal focusing (IF) lenses (mostly expensive long focal length > lenses) and lenses with close range correction by moving groups of > elements (mostly wide angle retrofocus lenses such as the Nikkor > 2.8/24mm). Most lens design clues today are in getting silent, small, with no distortion and luminous lenses. That's secondary for me. I'm even thinking in triying to design my own lens and building it but I'm pretty sure that I will be reinventing wheel and would take a lot of work for me. > IMHO, the best way to ensure that the lens focuses entirely by moving > the optical assembly is to use something like a bellows unit for > focusing. and a suitable lens, possibly one without a focusing mount > (such as the long discontinued 100mm Bellows-Nikkor) Probably that is not a very practical solution, because lens should be compact and with as few mobile parts as possible. But a very simple lens with just too groups would work fine. > As a side issue I'm somewhat intrigued by the assertion that the > Nikkor 1.8/50mm changes focal length by 0.2mm over its focusing range. > Where does that number come from and is that figure really larger than > the measurement error? I wrote 0.2 mm as an aproximate typical value in my projects, not over all his focusing range, but from the usual distances, say 5 meters, to near infinity. I can calculate focal length of the lens using calibration cards with known sizes and with lots of ligth. The exact focal length value is not known, but the ratio focal length / format size, being format size fixed. Focal length should be estimated at each picture with a 1/5000 error, but can change more between pictures to some unknown extent because bundle adjustment method could recalculate it, but the more variation the more chances for bundle not to work. In my experience a variation of until 1/100 is tolerable, but should also be reduced if possible to improve final results accuracy. The lens has a limited depth of field. But with enough ligth I can close diafragm a lot and focus at hyperfocal distance so when I take a picture all object are in focus and measure image sizes of known objects at known distances to check how ratio varies (acounting for distortion) I'm thinking that another aproach to solve my problem is to look for fixed lenses with small hyperfocal distance and lock focus ring forever, but will need lots of ligth... Thank you for your attention. >The ideal lens for this is a pinhole, but is too impractical, you need >too much light and exposure time. I need a real lens that resembles a [quoted text clipped - 6 lines] >calibration, but the final results are better when this parameters change >the little. For a photographic lens with unit focusing the focal length does not depend on the object distance. Perhaps you are worried about the field of view instead? SignatureRose-colored glasses are never made in bifocals. Nobody wants to read the small print in dreams. Ann Landers [... big snip ...] >> As a side issue I'm somewhat intrigued by the assertion that the >> Nikkor 1.8/50mm changes focal length by 0.2mm over its focusing range. [quoted text clipped - 7 lines] >value is not known, but the ratio focal length / format size, being >format size fixed. That does not in any way convince me regarding you alleged accuracy. You are speaking of a 0.4% accuracy (i.e. 0.2mm of 50mm) in a quantity which is not easily measured directly. You really should do an error analysis of your measurements. Very small errors in the estimation of your object distance or your image magnification could be manifested as major errors in your final estimate. Come to think of it, even a very small error in the estimation of the locations of the nodal points of the optical system could result in significant errors in the estimation of the focal length. To be quite honest, I consider that possibility the most likely explanation of the apparent focal length change you claim to have to observed. While I do realize that there are effects which would show up as focal length changes with working distance (spherical aberration being the prime candidate), I do not know of any optical systems which would not potentially display such behaviour at any aperture larger than the diffraction limited aperture. I can not envision any complex lens system which would perform better at the sub 1% error range without a serious (and expensive) calibration and verification procedure. Finally, in any realistic measurement system, the issue of fault tolerance has to be considered. If your measurement system is so sensitive to errors that an error of less than 1% in an (essentially) invariant parameter is significant, it may not be practical. Signature- Helge Nareid Nordmann i utlendighet, Aberdeen, Scotland For e-mail, please use my initials with the same domain as the posted address > That does not in any way convince me regarding you alleged accuracy. Well, I have no reasons for claiming what is not true. Let me note that there migth be diferent accuracies we are talking about, see later. > You are speaking of a 0.4% accuracy (i.e. 0.2mm of 50mm) in a quantity > which is not easily measured directly. You really should do an error > analysis of your measurements. Very small errors in the estimation of > your object distance or your image magnification could be manifested > as major errors in your final estimate. Come to think of it, even a Worst, I wrote about a 0.02% accuracy... Believe me I did error all error analysis. My fault if not. Focal is not so critic. It is quite stable, in fact it is almost linear. It is almost directly related to the relative size of objects in image plane to their real size and distance, modelling distortion apart. A simple relative size measurement with subpixel calibration marks gives an accuracy in image size of at about 1/10000, and a handheld laser meter gives you a 1/10000 too in real sizes. There are some practical difficulties to get to this accuracy sometimes and requires statistical treatment to do oversampling, but there is no big problem, in some projects could get to 1/20.000. 1/5000 is rutine, 1/10000 requires some care and 1/20000 is hard but possible. Note: All measurements are _relative_, ie, adimensional or angular if you prefer. > very small error in the estimation of the locations of the nodal > points of the optical system could result in significant errors in the > estimation of the focal length. To be quite honest, I consider that You don't need to know exactly where nodal points of every lens are to calculate its empirical focal length. You can explicitally write focal length from sizes and distances of objects and images, being the mayor problem to know exact absolute image size (CCD size in mm) But I didn't claim to measure focal length so accuratelly (in fact I claim nothing), but how it changes between photos, Don't know focal length absolute value exactly, but focal legth / sensor size ratio, ie, angle, a relative measurement. That 0.2 mm change is very acurate by far in the accuracy range I have, I could say it changes 0.24 mm but I dont claim to say it is 50.24 mm. The key is I don't know the exact CCD size, but doesn't matter. > possibility the most likely explanation of the apparent focal length > change you claim to have to observed. I don't know why it changes, just measure it. I can even tell you how much the principal point of the lens moves and distortion parameters. A note: -That 0.2mm change is the tipical expected between two consecutive photos taken with the lens locked, that is, without moving focus. Sometimes this value is even smaller and can be unnoticeable. -When refocusing and trying to set the focus in the same position one day after, looking at marks tolerace grows to at least 1mm variation in focal length. -In "autofocus" mode, that is, moving around the object in a real situation and allowing focus changes between 5 to infinity meters to the rigth distance could be 2-3 mm and at all ranges becomes 5 mm (1%) -But every "sample" of focus length taken from calibration photographs is at least 1/5000 typical accuracy. > While I do realize that there are effects which would show up as focal > length changes with working distance (spherical aberration being the > prime candidate), I do not know of any optical systems which would not As I understand it, spherical aberration causes defocusing. Simply stated: there is not noticeable defocusing in the lenses I use, when well used, this is easy to see in photos. Even if there is some, the image of a dot becomes a slightly blurred dot, but its center it's at the same image point. It has no metric impact and only a ligth impact in accuracy for subpixel methods. If you mean distortion as a form of aberration, yes, barrel distortion changes image size, but it is calibrated and well known for correction before calculating focal length. > potentially display such behaviour at any aperture larger than the > diffraction limited aperture. I can not envision any complex lens > system which would perform better at the sub 1% error range without a > serious (and expensive) calibration and verification procedure. It depends of what value of the lens are you talking about. A (comercial) lens is very accurate at forming images. Take note that to focus every pixel in a 12 Mpixel camera requieres some kind of fine accuracy. The ray_angle_to_pixel function migth not be linear and change bewteen photos, but it is quite accurate in a single photo when modelled correctly (less than 20 numerical parameters to model). I don't need a lens with 50.323456 mm focal legth, I need any lens, that doesn't move (much) its focal length, not in a year lapse, just between two consecutive takes or one day to another and even when changing focusing. How "much" ? as little as possible. 0.4% for fixed focus (playing with hyperfocal distance) and 4-6% for "autofocus" variations works fine as a reference, 20-25% variations as in some IF lens are not acceptable. Focusing fixed at hyperfocal distance works fine with most lenses, but it is not always practical (note that pictures are taken outdoors and sometimes from cranes or even helicopter) > Finally, in any realistic measurement system, the issue of fault > tolerance has to be considered. If your measurement system is so > sensitive to errors that an error of less than 1% in an (essentially) > invariant parameter is significant, it may not be practical. Some background: I was using and working on this measurement system for 5 years now to measure real objects and I can asure you that even with some dificulties (like the fact that some lenses do not behave as I would like) measures obtained by this method (a few thousands per project) fit with real measures. It is already practical, what I try is to improve it. Ray proyection and intersections are correct, for sure, and that means that modelled camera parameters are right too. The influence of every parameter in solutions is of geometric nature and easy to analize. Nothing new, its a photogrametric software tool. It can tolerate any change in focal length, but the more variation, the more uncertainity and more redundant measurements are needed. It should figure as an unknown parameter with instead of a known parameter. The numerical method is a fitting one with bundle adjustment. So if you want a fast solution it is a must to give some range estimations for the unknown values, the narrower the better. I know, for example, that my CCD heigth to width ratio is fixed, this is quite obvious and is a very important optical parameter to know. ´ Would be nice to being able to narrow the focal length range too, there is no definite goals, any improvement would work. Does this give a better understanding of the problem? Thanks for your patience. >> As a side issue I'm somewhat intrigued by the assertion that the >> Nikkor 1.8/50mm changes focal length by 0.2mm over its focusing [quoted text clipped - 4 lines] > all his focusing range, but from the usual distances, say 5 meters, to > near infinity. I can calculate focal length of the lens using Wrong! My fault, I was confusing the thread... sorry Above should read " 2 mm " or maybe even 4mm, instead of "0.2mm" 0.2mm is the typical change when focus is locked. <snip> > As a side issue I'm somewhat intrigued by the assertion that the > Nikkor 1.8/50mm changes focal length by 0.2mm over its focusing range. > Where does that number come from and is that figure really larger than > the measurement error? If none of the lens separations change, as I believe is the case for this lens, then of course true paraxial focal length does not change. Measuring focal length of a wide-ish field lens with significant uncorrected aberrations is a non-trivial exercise. I would guess that the original poster is measuring a calibration value which factors in distortion and defocus to balance spherical aberration, field curvature etc. In this case a 0.2mm variation with focus is perfectly believable. Robin. ><snip> >> As a side issue I'm somewhat intrigued by the assertion that the [quoted text clipped - 6 lines] >Measuring focal length of a wide-ish field lens with significant >uncorrected aberrations is a non-trivial exercise. I am in complete agreement with the above, which is why I query the accuracy of an alleged measurement of 0.4% deviation over the operational range. If valid, it is a pretty impressive measurement of a non-trivial quantity. >I would guess that the original poster is measuring a calibration >value which factors in distortion and defocus to balance spherical >aberration, field curvature etc. In this case a 0.2mm variation with >focus is perfectly believable. I would agree that all of those parameters are _somehow_ included in the measurements, but are they really properly calibrated and accounted for? My guess is that the 0.2mm variation - _if_ accurately measured - is a consequence of the above factors, with no real analysis being applied to where the variation is coming from. This is relevant regardless of which lens is being used for the final application, since those factors are going to be present for any lens system with an aperture larger than a diffraction limited pinhole. Signature- Helge Nareid Nordmann i utlendighet, Aberdeen, Scotland For e-mail, please use my initials with the same domain as the posted address > ><snip> > >> As a side issue I'm somewhat intrigued by the assertion that the [quoted text clipped - 25 lines] > application, since those factors are going to be present for any lens > system with an aperture larger than a diffraction limited pinhole. Even with pinholes you need to define the mechanical back focus to get meaningful repeatability. Nailing down all the variables is not easy, and even if you do that I have to wonder how useful a true paraxial focal length value is in practical terms. A practical calibration method that I have frequently used is to measure loads of image heights against angles (or object height for a relay) and fit the resulting data set to the function: h=f tan(theta) + a tan^3(theta) + b tan^5(theta) etc. (more or less terms depending upon the likely complexity of the distortion function) The value of f falls straight out of this and you can extract distortion values at any angle from: d=[a tan^2(theta) + b tan^4(theta) etc.]/f You can also find distortion maxima, and the field angles at which they occur, by taking the first differential of this. This is all very useful for most practical purposes, but how close is f to a true paraxial value? Unless the lens is strictly telecentric, then that depends upon the choice of focal plane. Robin. > Any recomendation? Any website or catalog where I can check this > caracteristics or any clue o keyword to look for. Bob May had it right. Try rec.photo.digital, or try... http://www.stevesforums.com/forums/view_forum.php?id=16 for Fuji discussion group http://www.stevesforums.com/forums/view_forum.php?id=81 for discussion group that specializes in architectural photography Optiker |
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.