Olsen Posted June 9, 2011 Share #121 Posted June 9, 2011 Advertisement (gone after registration) If the IFP is caused or influenced by camera-to-camera variation in the position of the microlens array in relation to the sensor, it cannot be fixed with a firmware update that necessarily applies the same correction to every camera. But it could in principle be tackled by giving each camera custom firmware tailored to its particular sensor. In practice this might be done by splitting the firmware code in two: one part with the IF correction data and the other with everything else. The first part would only be updated if the sensor was changed; the second would be field-updated as normal. Whether the M9 hardware allows this is another question. Giordano, Thanks, That means that Leica will not issue an up date that will solve this 'once and for all'. 1) My M9 produces Italian Flag only on my VC 15 (old design without RF coupling). My other lenses are 'perfect': WATE, 35 mm Cron, 50 mmm Noctilux 1,0 etc. So, I can't 'storm down' to Solms and claim it fixed, can I? I will change lens adaptor (I have two) on the VC 15 and see if it helps. Still I can regard myself to be lucky: My individual M9-camera can't be that bad. 2) I did some parallel tests with my two cameras (M9 & 1Ds III) last winter on a bright sunny day with a lot of snow.with different lenses. While my M9/WATE combo produced no Italian Flag whatsoever, my 1Ds III 16-35 mmL II did produce an Italian Flag effect, though faint.... 3) There is supposed to be a Leica R lens that many modify to mount on Canon EOS cameras. Because this lens produces far less vignetting than the Canon wide angle lenses. - Vignetting can be several stops from corner to the center of the pictures on most DSLRs today. This particular R-lens must be a favoured design for future digital cameras. Or? 4) I haven't come around to installing a C1 6 yet. My wife is all into IPhone and Mac Air. I am the only one fumbling around with PC in the house. I am planning to buy a 17" portable Mac. I have never become familiar with Lightroom, so C1 6 will be just fine with me. Link to post Share on other sites More sharing options...
Advertisement Posted June 9, 2011 Posted June 9, 2011 Hi Olsen, Take a look here Towards an explanation of the Italian Flag Phenomenon. I'm sure you'll find what you were looking for!
pgk Posted June 9, 2011 Share #122 Posted June 9, 2011 Forgive me if this has been suggested already (I HAVE looked) but assuming that the IR filter glass is secured into the camera in a standard and repeatable way (ie assembled in the same sequence), could it be that stresses on this glass lead to a 'somewhat' predictable, very slight bowing of the filter glass itself, which combined with very oblique light from wide-angles, could lead to problems or could at least add to other problems (my experience in problem solving suggests that few problems have one simple cause)? or could it be that a very slight non-parallelness of this cover glass could again lead to problems? My guess is that tolerances are very, very tight indeed and I just wonder if a known bias in production might result in an easier to adjust/correct for problem than trying to achieve absolute perfection in production? Link to post Share on other sites More sharing options...
Lindolfi Posted June 9, 2011 Author Share #123 Posted June 9, 2011 But that could mean that IFP has nothing to do with micro-lenses and/or their chromatic aberration. As described by PhaseOne, it evokes the thought of a sort of prisma effect in the glass covering the sensor No, that does not work, because any dispersion of a ray of white light will be complemented by colours from neigbouring rays ( that also disperse) to form white light. Only at the edge of the cover glass you may see colour dispersion. The same holds for the suggestion of pgk about mounting of the cover glass. Link to post Share on other sites More sharing options...
ho_co Posted June 9, 2011 Share #124 Posted June 9, 2011 ... it depends mainly on how short the distance is between the rear lens of your type of 21 and the sensorplane ... To be technical, I think in fact it's not how close the rear element is to the sensor, but how close the exit pupil is. Link to post Share on other sites More sharing options...
Olsen Posted June 9, 2011 Share #125 Posted June 9, 2011 To be technical, I think in fact it's not how close the rear element is to the sensor, but how close the exit pupil is. What is the exit pupil? How do you measure it's 'closeness' to the sensor? Link to post Share on other sites More sharing options...
mjh Posted June 9, 2011 Share #126 Posted June 9, 2011 What is the exit pupil? How do you measure it's 'closeness' to the sensor? The exit pupil is the virtual image of the aperture as seen through the rear element. It’s where the rays hitting the sensor appear to originate so if you know about the position and size of the exit pupil you know about the incident angles. For example, a telecentric lens is one where the exit pupil is at infinity so all the principal rays would hit the sensor at an incident angle of 0°. Link to post Share on other sites More sharing options...
andyb Posted June 9, 2011 Share #127 Posted June 9, 2011 Advertisement (gone after registration) OT, but maybe of interest. The exit pupil is important while choosing the right eyepieces for your scope in visual astronomy (Dobsons and so on...). The diameter of the exit pupil of the eyepiece shouldn't be bigger than your eyes max. pupil diameter at night (i.e. approx 6-7 mm in the adult), otherwise the eye doesn't catch all the light gathered from the main mirror. It is calculated in dividing the focal length of the eyepiece by the f number of the main mirror. All the best, Andreas Link to post Share on other sites More sharing options...
giordano Posted June 9, 2011 Share #128 Posted June 9, 2011 Maybe some tests with a pinhole would be worthwhile. At least that way the size and position of the exit pupil are easy to establish. Link to post Share on other sites More sharing options...
Lindolfi Posted June 10, 2011 Author Share #129 Posted June 10, 2011 Yes I agree about simplifying the ray bundles onto the sensor using a pinhole. Also, to be able to reject my explanation of the diagonally shifted microlenses, I'm setting up a photo-microscope assembly to take pictures from the centre and sides of the M9 sensor seen straight from above. Presently I have a setup that makes a single microlens cover an array of 15x15 pixels in the camera used for the micro photo. That means that possible shifts of the microlenses relative to the colour sites must me visible. But because of the value of the subject at hand (the M9) I'm still working on the safety precautions. Link to post Share on other sites More sharing options...
otto.f Posted June 10, 2011 Share #130 Posted June 10, 2011 Good work!! Link to post Share on other sites More sharing options...
Lindolfi Posted June 10, 2011 Author Share #131 Posted June 10, 2011 OK, here is the first image from the center of the M9 sensor. The width of the image is 340 micron (so less than 0.01 % of the surface of the sensor). The big problem of photographing the sensor with a microscope is that you have no room for lighting in that narrow M9 house when you use a high magnification apo microscope objective because it is so long. So I ended up using a Zeiss Luminar 40/4.5 which has enough clearance for lighting, but not very high magnification. The safety precautions consisted of limiting the microscope travel downward, so that the microscope objective could not touch the sensor. Link to post Share on other sites More sharing options...
Lindolfi Posted June 11, 2011 Author Share #132 Posted June 11, 2011 Have improved the microscope setup with a better objective with twice the resolving power, but I can not make identical images from the center and the four corners of the sensor due to differences in lighting. When I shift the objective, I also have to shift the lighting. Since the layer of microlenses is both reflective and translucent and is 3D in shape, I can not get identical conditions to be able to compare the four corners with the center of the sensor. The solution is to throw the light through the objective itself, so that it comes straight from above onto the sensor and the lighting conditions become identical. So that is the next step. For now, all images from the center of the sensor that I have obtained show that the microlenses are within 1.5 micron aligned with the Bayer pattern of colour wells. The colour well (and pixel) pitch is 6.8 micron. So that points in the direction of a rejection of my diagonal microlens layer shift explanation. We need better images. Link to post Share on other sites More sharing options...
otto.f Posted June 11, 2011 Share #133 Posted June 11, 2011 Appreciate this very much! I check my iPad a few times a day if there's news from the Lindofli vs. IFP front Link to post Share on other sites More sharing options...
giordano Posted June 11, 2011 Share #134 Posted June 11, 2011 For now, all images from the center of the sensor that I have obtained show that the microlenses are within 1.5 micron aligned with the Bayer pattern of colour wells. The colour well (and pixel) pitch is 6.8 micron. So that points in the direction of a rejection of my diagonal microlens layer shift explanation. We need better images. I don't think this militates against the theory that variations in the relative position of the microlens array and the sensor are involved in the IFP. With the shifted microlenses, the pitch of the microlens array may not be constant and is certainly not a constant 6.8μm. At the short edges of the array the offset between the centres of a colour well and the corresponding microlens is of the same order of magnitude as the radius of the lens - so I feel that ±1.5μm isn't accurate enough. But I can't imagine how to increase the magnification tenfold without dismantling the camera. In fact in Leica's and Kodak's position - with full knowledge of the dimensions of all components of the sensor pack - the best way to check for accurate positioning of the microlens array might be to make an image with a pinhole and look for traces of Italian flag in the raw (not RAW) sensor output. Link to post Share on other sites More sharing options...
Lindolfi Posted June 11, 2011 Author Share #135 Posted June 11, 2011 To make my statement clearer: I took images only of the center of the sensor and in the center the shift is less than 1.5 micron. The radius of the microlenses is about 3 micron. The explanation I gave demands that there is a constant shift of the layer of microlenses over the whole sensor in the direction from the left bottom corner to right top corner diagonal (of the image) on top of the shift purposely introduced to help the wide angle lenses. I agree that we need better images and think that I can only get to about 0.6 micron resolution by improving my lighting and using a 60/0.85 objective without oil immersion. The suggestion to get to the raw data is fine as long as it is a realistic goal. Anyone? On the other hand: in principle we can back calculate the raw data (values from each colour well) from the RAW file (RGB values per pixel) by inverting the Bayer algorithm: no information is lost. In fact the Bayer forward algorithm produces three times more data than captured. (That is why you get these crazy colour effects when photographing black and white gratings close to the Nyquist frequency of the sensor without an AA filter). So it should be possible to do it. Link to post Share on other sites More sharing options...
otto.f Posted June 11, 2011 Share #136 Posted June 11, 2011 the best way to check for accurate positioning of the microlens array might be to make an image with a pinhole and look for traces of Italian flag in the raw (not RAW) sensor output. If the exit pupil of the pinhole comes close enough to provoke IFP Link to post Share on other sites More sharing options...
Lindolfi Posted June 12, 2011 Author Share #137 Posted June 12, 2011 Here is an Italian Flag Phenomenon image made with a pinhole mounted on the M9 located at a distance of 28 mm (on the flange) and a diameter of 0.2 mm. In front of the hole there was the opaque piece of non-shiny plastic to create non-directional light with the colour of the ambient light. In the post process, only the very center of the image was used for white balance, so that the colour of the ambient light and the proposed setting of LightRoom (coming from the M9) were overruled. and here is the colour analysis of the image along the left-bottom to right-top diagonal Here is the image through the same pinhole at 69 mm distance from the sensor (on the flange of the visoflex house with raised mirror) an the colour analysis Link to post Share on other sites More sharing options...
otto.f Posted June 12, 2011 Share #138 Posted June 12, 2011 [/img] So this would be in favor of the diagonal hypothesis? Link to post Share on other sites More sharing options...
Lindolfi Posted June 12, 2011 Author Share #139 Posted June 12, 2011 So this would be in favor of the diagonal hypothesis? No it is not. The maximal discolouration (at least in my M9) is maximal along the vertical axis of the image. To demonstrate this below the cyan-magenta plot along the largest centered circle in the image in degrees starting at 3 o'clock. You can see that maximal cyan is at 12 o'clock (90 degrees) and maximal magenta is at 6 o'clock (270 degrees). To show that you get partially fooled by the total image, I have cut out the central circle, leaving the rest 50% grey: It is obvious that magenta cast is in the lower part of the circle and cyan cast in the upper part. So what is now needed is the same image from other M9 camera's: a pinhole located on the flange of the camera with a diameter of 0.2 mm and covered by white opaque non-shiny plastic illuminated with diffuse ambient neutral light (like from the clouded sky). This would show if the gradient of magenta to cyan is always in the same direction. Link to post Share on other sites More sharing options...
t024484 Posted June 12, 2011 Share #140 Posted June 12, 2011 This is what my M9 does. Magenta below and cyan above. Hans Welcome, dear visitor! As registered member you'd see an image here… Simply register for free here – We are always happy to welcome new members! Link to post Share on other sites Simply register for free here – We are always happy to welcome new members! ' data-webShareUrl='https://www.l-camera-forum.com/topic/153464-towards-an-explanation-of-the-italian-flag-phenomenon/?do=findComment&comment=1699688'>More sharing options...
Recommended Posts
Archived
This topic is now archived and is closed to further replies.