On 1/10/2012 11:18 AM, glen herrmannsfeldt wrote:> Clay<clay@claysturner.com> wrote: > > (snip) >> Conversely this is why some telescopes work very poorly for imaging >> nearby subjects. I have a 6 inch Ritchey Chretien telescope which if >> used to image something as close as 200 meters works very poorly >> whereas for astronomical objects it works extremely well. > >> I recall in one of the photography forums discussion of Nikon's famed >> 200-400mm F/4 zoom not being sharp at inifinity. It is pretty sharp >> there but it is much sharper at 25 to 50 meters. It could have been >> engineered to be sharp at infinity, but that wasn't its main design >> purpose. Now with high resolution digital sensors, people are better >> able to test their optical systems and see their design compromises. > > I have wondered how to compare high-resolution digital sensors > to ultra-fine grain film. Film, like many analog systems, has a tail > on the resolution curve that slowly falls of, where digital has a > sharp drop. Well, as sharp as the low-pass filter allows.What low-pass filter? That's interesting. Few lenses nowadays have resolutions good enough to tax the best digital sensors, and those are in color. The best available lenses (we used Wray f/4 diffraction limited lenses) on B&W Kodak high-res plate on which the emulsion had been dyed black to keep the image entirely on the surface (depth of focus, and all that) routinely recorded 1000 line pairs/mm at 50% contrast. That's a lot of pixel equivalents on a 2'x2' plate. When they make digital sensors with 25e8 pixels, let me know! Jerry -- Engineering is the art of making what you want from things you can get. �����������������������������������������������������������������������
Internet access during exams
Started by ●January 5, 2012
Reply by ●January 10, 20122012-01-10
Reply by ●January 10, 20122012-01-10
On Jan 10, 3:11�pm, Jerry Avins <j...@ieee.org> wrote:> On 1/10/2012 11:18 AM, glen herrmannsfeldt wrote: > > > > > > > Clay<c...@claysturner.com> �wrote: > > > (snip) > >> Conversely this is why some telescopes work very poorly for imaging > >> nearby subjects. I have a 6 inch Ritchey Chretien telescope which if > >> used to image something as close as 200 meters works very poorly > >> whereas for astronomical objects it works extremely well. > > >> I recall in one of the photography forums discussion of Nikon's famed > >> 200-400mm F/4 zoom not being sharp at inifinity. It is pretty sharp > >> there but it is much sharper at 25 to 50 meters. It could have been > >> engineered to be sharp at infinity, but that wasn't its main design > >> purpose. Now with high resolution digital sensors, people are better > >> able to test their optical systems and see their design compromises. > > > I have wondered how to compare high-resolution digital sensors > > to ultra-fine grain film. Film, like many analog systems, has a tail > > on the resolution curve that slowly falls of, where digital has a > > sharp drop. Well, as sharp as the low-pass filter allows. > > What low-pass filter? That's interesting. > > Few lenses nowadays have resolutions good enough to tax the best digital > sensors, and those are in color. The best available lenses (we used Wray > f/4 diffraction limited lenses) on B&W Kodak high-res plate on which the > emulsion had been dyed black to keep the image entirely on the surface > (depth of focus, and all that) routinely recorded 1000 line pairs/mm at > 50% contrast. That's a lot of pixel equivalents on a 2'x2' plate. When > they make digital sensors with 25e8 pixels, let me know! > > Jerry > -- > Engineering is the art of making what you want from things you can get.- Hide quoted text - > > - Show quoted text -In most digital cameras a low pass filter is either made part of the sensor or like in a lot of point and shoot cameras effected by the lens which becomes the low pass filter. Although in some cases the filter was omitted on purpose and yet sometimes the lens out resolved the sensor and nasty things like color moire' patterns showed up in the image. An example of such a camera was the Kodak Pro 14N - a 24 by 36mm sensor with 14MP back in 2003! Certainly for high res film, just use holographic film such as Agfa 8E75HD which resolves to half a wavelength of red light. Kodak had a spectroscopic emulsion (649F - IIRC) which was almost as good. But these films are much too slow for use in hand held cameras. I've tried! I think the sharpest lenses these days are made by Nikon for use in their step and repeat machines for etching silicon wafers. But these are special purpose and would not likely do well to image a bird 50 feet away. Plus their field of coverage is limited. I think your big sensor with 25E8 pixels will become reality sooner than you think! Kodak makes a 50MP chip now that is used in the Hasselblad H3D II camera. Plus if you have a piece of film with 25E8 pixels of info in it, how do you get it out? Slides use to be read with drum scanners and oil immersed lenses. Kodak Chip: www.kodak.com/ek/US/en/KAF-50100ProductSummary.htm Clay
Reply by ●January 10, 20122012-01-10
Jerry Avins <jya@ieee.org> wrote: (snip, I wrote)>> I have wondered how to compare high-resolution digital sensors >> to ultra-fine grain film. Film, like many analog systems, has a tail >> on the resolution curve that slowly falls of, where digital has a >> sharp drop. Well, as sharp as the low-pass filter allows.> What low-pass filter? That's interesting.The descriptions of DSLR's usually indicate a low-pass filter. Among others, you get dust on the LPF, not on the image sensor. It might be that point-and-shoot digital cameras don't have/need one, but DSLR's, at least Nikon's, do.> Few lenses nowadays have resolutions good enough to tax the best digital > sensors, and those are in color. The best available lenses (we used Wray > f/4 diffraction limited lenses) on B&W Kodak high-res plate on which the > emulsion had been dyed black to keep the image entirely on the surface > (depth of focus, and all that) routinely recorded 1000 line pairs/mm at > 50% contrast. That's a lot of pixel equivalents on a 2'x2' plate. When > they make digital sensors with 25e8 pixels, let me know!I have some Kodak TP2415 Technical Pan film which seems to be rated at only 100 lp/mm at 50%, with the graph extending on to 200 lp/mm at 30%. Some years ago, I remember using TP2415 for lens resolution test experiments in an optics class. I believe with resolution test targets from NBS (they were old, from the pre-NIST days). One suggestion for focus was to take a few slightly on either side of what you thought was the correct focus, in case it was a little off. There are articles on lens resolution tests with DSLR's, it seems that the sensors are high enough resolution to make it a useful thing to do. -- glen
Reply by ●January 10, 20122012-01-10
Clay <clay@claysturner.com> wrote: (snip, I wrote)>> > I have wondered how to compare high-resolution digital sensors >> > to ultra-fine grain film. Film, like many analog systems, has a tail >> > on the resolution curve that slowly falls of, where digital has a >> > sharp drop. Well, as sharp as the low-pass filter allows.>> What low-pass filter? That's interesting.>> Few lenses nowadays have resolutions good enough to tax the best digital >> sensors, and those are in color.(snip)> In most digital cameras a low pass filter is either made part of the > sensor or like in a lot of point and shoot cameras effected by the > lens which becomes the low pass filter.> Although in some cases the filter was omitted on purpose and yet > sometimes the lens out resolved the sensor and nasty things like color > moire' patterns showed up in the image. An example of such a camera > was the Kodak Pro 14N - a 24 by 36mm sensor with 14MP back in 2003!> Certainly for high res film, just use holographic film such as Agfa > 8E75HD which resolves to half a wavelength of red light. Kodak had a > spectroscopic emulsion (649F - IIRC) which was almost as good.I remember the 649F as glass plates. I do remember a 35mm holographic film. We used it in an optics experiment to make holographic diffraction gratings. That would have been 1978, if that helps any.> But these films are much too slow for use in hand held cameras. > I've tried!Did you try with image stabilizing lenses? That gets you a few more stops much of the time.> I think the sharpest lenses these days are made by Nikon for use in > their step and repeat machines for etching silicon wafers. But these > are special purpose and would not likely do well to image a bird 50 > feet away. Plus their field of coverage is limited.> I think your big sensor with 25E8 pixels will become reality sooner > than you think! Kodak makes a 50MP chip now that is used in the > Hasselblad H3D II camera.> Plus if you have a piece of film with 25E8 pixels of info in it, how > do you get it out? Slides use to be read with drum scanners and oil > immersed lenses.The lens test papers I was reading suggested that a microscope was the only way to read the results off the film. -- glen
Reply by ●January 11, 20122012-01-11
> > The lens test papers I was reading suggested that a microscope was > the only way to read the results off the film. > > -- glenI've certainly put negatives under a microscope to see what was really going on. Drum scanners used a glass cyclinder around which the film was tightly wrapped. Inside of the cyclinder was the light source and outside was effectively a microscope with a photocell (photomultiplier tube). Oil was used both for the immersion objective and to avoid Fizeau fringes between the film and glass. The microscope/photodetector was on a screw drive that progressed along as the drum rotated - sort of like an old Edison recording device. There used to be Service Bureaus who would scan your film for you for around $25/scan. I had a few done that way back in the day before film scanners become reasonably priced. I was only in the shoot film and "scan it" mode of imaging for a few years until direct digital capture become as good or better than using film. Even though I miss the time spent in a darkroom, I don't miss the aggravation associated with shooting film. But back to one of your early questions, one of my scanners would do 2700 color pixels per inch (~100/mm) and this seemed to get most of the info from iso 100 film. I shot both C-41 and E-6 based films and they both were well scanned at this resolution. My old Kodachrome 25 stuff wouldn't scan as well since the red dyes were different plus the scanner would also scan with IR and use that information to remove fingerprints and dust. If you turned off the dirt removal by disabling the IR scan, Kodachrome could be scanned although the red channel was still a bit noisey. Slides are notoriously contrasty, aggravated by the practice of slightly underexposing them to increase saturation, and their high optical density would challenge many of the early home/ semipro film scanners. C-41 negatives on the other hand were quite easy to scan. One of the things that has come out of the question "how much info is in a picture?", is the realization that only parts of the image have high info density while others don't. One interesting technique in imaging being used now is tone mapping. This is really needed when doing HDR (high dynamic range) imaging since the source's dynamic range far exceeds that of the display device. So the tone mapping may be used to compress an image's dynamic range, but it can do much more. There are some tone map algos that will try to do things like make each brighness level occur nearly equally often and this may be done globally or be confined to a local regions of the image. If pushed too far the image takes on a painterly or cartoonish look. I find the tonemapping useful even with a single (non combined HDR) image. It is just another tool in the box. Even processes like "unsharp masking" that improve the apparent image detail exploit some human visual perceptual factors including edge detection by working on amplifying local contrast. I like working with images in a LAB color space since the luminance (L) is separated from the two color channels (A & B) and then color manipulations become quite easy. Clay
Reply by ●January 11, 20122012-01-11
Clay <clay@claysturner.com> wrote: (snip)> But back to one of your early questions, one of my scanners would do > 2700 color pixels per inch (~100/mm) and this seemed to get most of > the info from iso 100 film. I shot both C-41 and E-6 based films and > they both were well scanned at this resolution. My old Kodachrome 25 > stuff wouldn't scan as well since the red dyes were different plus the > scanner would also scan with IR and use that information to remove > fingerprints and dust. If you turned off the dirt removal by disabling > the IR scan, Kodachrome could be scanned although the red channel was > still a bit noisey. Slides are notoriously contrasty, aggravated by > the practice of slightly underexposing them to increase saturation, > and their high optical density would challenge many of the early home/ > semipro film scanners. C-41 negatives on the other hand were quite > easy to scan.Easy to scan, hard to print. C41 (and, as far as I know, C22) are what would be called low-contrast in B&W photography. Intensity changes are compressed (in log space) by about a factor of two, which is reversed in printing. Among others, when printing you decide the color change needed, looking through the filter, and then change half that much. (Another factor of two if you put the filter on top of the print.) Also, for the exposure you have to be more accurate than would be expected with normal B&W printing. Printing E6, everything works out as you would normally expect. -- glen
