Re: How to photograph an oil painting

Discussion in 'Digital Photography' started by Wolfgang Weisselberg, Apr 2, 2013.

  1. Jennifer Murphy <> wrote:
    > On Fri, 29 Mar 2013 19:28:47 -0700, Jennifer Murphy


    > I now have some photos of the painting. They were taken by a friend with
    > a Nikon DSLR camera. But now I think we may have taken the photos at the
    > wrong resolution. (sigh)


    > The raw psd file is 130 MB. I opened it in IrfanView and brought up the
    > image properties panel. It reports these properties:


    > Compression: Run length encoding


    irrelevant

    > Resolution: 240 x 240 DPI


    irrelevant and trivial to change

    > Original size: 4256 x 2832 Pixels (12.05 MPixels) (3:2)
    > Current size: 4256 x 2832 Pixels (12.05 MPixels) (3:2)


    Relevant

    > Print size (from DPI): 45.0 x 30.0 cm; 17.7 x 11.8 inches


    Changes when you change "DPI"

    > Original colors: 16,7 Millions (24 BitsPerPixel)
    > Current colors: 16,7 Millions (24 BitsPerPixel)


    Relevant

    > Number of unique colors: 597146


    Depends on the image

    > Disk size: 127.81 MB (134,020,067 Bytes)
    > Current memory size: 34.48 MB (36,159,016 Bytes)


    Quite irrelevant.

    > I asked my friend to take the photo at 300 DPI at the full size of the
    > painting, which is 24.25 x 20.25 inches.


    Does the camera of the friend have at least 7275x6075 pixels?
    That will come out to 6075x9113 pixels due to the 2:3 format
    for DSLRs, and means 55 MPix --- which in turn means you need to
    buy your friend a medium format camera with digital back first!

    Alternatively you need to make exposures of different parts
    of the image and stitch them together. This is however even
    more complicated.

    > This appears to be only 240 DPI
    > at 17.1 x 11.8 inches and the portrait doesn't fill the image. If I draw
    > a selection box around the portrait, it's only 2885 x 2401 of the
    > overall 4256 x 2832 for the whole image.


    2401 to 2832 is 15%. Get the camera a bit closer to the painting.

    > I asked her to use the perspective tool in Photoshop to square up the
    > image, which she did. I got another PSD file and a JPG. The PSD file is
    > about 83 MB and the properties information is:


    > This image is just the portrait, no border, but it's only 9.7 x 8.1 at
    > 300 DPI. I have to set the DPI down to about 120 x 120 to get a print
    > size close to 24 x 20.


    See above.

    > The JPG file is only 4.4 MB. It's properties are:


    > Compression: JPEG, quality: 98, subsampling OFF


    Quality 98 is very high, and 4.4 MB isn't a problem at all.
    (Run length encoding is very suboptimal for images, JPEG is
    very good and *is* after all lossy.)

    > To my eye on the screen, it looks identical to the PSD file, which is 20
    > times larger.


    That's the idea.


    > The problem I have is that I think I have reached the end of my friend's
    > patience, possibly the capacity of her camera, and maybe even the extent
    > of her photography skills.


    > Now what?


    Pay your friend. This is starting to become work, not fun
    --- and lots of work on the computer afte the shooting.

    As to camera --- see above.

    > I called the local camera store. They will rent me a Nikon D80 (I think
    > that was the model number)


    D800. 36 MPix. You need 55 MPix at 2:3 at your 300 "DPI"
    wish, see above. The D800 already has the highest MPix count
    of all FF DSLRs.

    > with the right lens for about $180 for the
    > weekend. That's not too bad, but I'm not sure I'll be able to make it
    > work. I'd hate to spend $180 and still not get a good photo.


    You need to work *very* carefully. Exact focussing (probably
    manual). Heavy tripod and remote release.

    Oh, and see above, you need to stitch multiple photos for
    300 "DPI" wish. Which means you need more gear (nodal point
    adapter, software) and the knowledge how to adjust and use them.

    > Maybe I'll look for another friend.


    You'll be wearing tje

    > This turned out to be a lot more difficult that I ever dreamed. Several
    > people I have talked to


    ?

    -Wolfgang
    Wolfgang Weisselberg, Apr 2, 2013
    #1
    1. Advertising

  2. Wolfgang Weisselberg

    nospam Guest

    In article <>, Jennifer
    Murphy <> wrote:

    > OK, I think I sorta understand. What matters is the total number of
    > pixels (X x Y). That's the raw data. If I have 3000x3000 pixels, I can
    > have a a 10"x10" print at 300 dpi, or a 20"x20" print at 150 dpi. I use
    > dpi here because I am talking about the dots on the printed page. Right?


    should be ppi but it's obvious what is meant. printers use multiple
    dots per pixel.

    > I snipped the rest because I'm not going to take multiple shots and
    > stitch them together (horrors!). I think you were joking with me anyway.


    he's not. it is an option, and not all that difficult. it's just time
    consuming.

    the stitching is done automatically by panorama software and you can
    get substantially higher resolution than with a single photo.

    > And I'm not going to go get a camera that I probably can't lift so I can
    > get a raw image at 300 ppi for a 20x24 inch image. As you and others
    > have said, I probably already have something that will be good enough.


    they're not that heavy. :) the problem is finding one and getting past
    the shock of how much it will cost to rent it.

    > However, I still have friends with a photography habit who are willing
    > to play around on this project. And I am learning a few things.


    always a good thing.

    > Another friend came over with a Canon DSLR. I forgot (again) to get the
    > model number, but it was huge. She just sent me two images.


    based on the data below, it's fairly old, probably a 1d. does she shoot
    sports? another possibility is it was set to a lower resolution.

    > One is the raw camera file. It has a .CR2 extension. Here are the
    > properties:
    >
    > Compression: None
    > Resolution: (Blank) ____ x ____ DPI
    > Original size: 2496 x 1664 Pixels (4.15 MPixels) (3:2)
    > Current size: 2496 x 1664 Pixels (4.15 MPixels) (3:2)
    > Print size (from DPI): 88.1 x 58.7 cm; 34.7 x 23.1 inches
    > Original colors: 16,7 Millions (24 BitsPerPixel)
    > Current colors: 16,7 Millions (24 BitsPerPixel)
    > Number of unique colors: 222026
    > Disk size: 12.92 MB (13,550,637 Bytes)
    > Current memory size: 11.88 MB (12,460,072 Bytes)
    >
    > I'm a little puzzled about the Print Size, because no Resolution was
    > specified. If I plug the pixel count (2496 x 1664) and the print size
    > into my handy little PPI calculator, it cranks out 72 PPI. And if I
    > enter 72 in the Resolution fields, the Print size stays the same.


    the resolution field means nothing, except to a few page layout apps.
    you can put anything you want in there.

    what matters is how many pixels you have and how big you print. that's
    what determines the actual ppi.

    > The Print size is about right for the actual size. The camera was a ways
    > back, so there is about 5" of border left and right and about 2" top and
    > bottom. So if we get the camera closer or zoom in (to avoid fish eye
    > distortion), we should be able to get the resolution a little higher. If
    > I plug the actual portrait size (24.25 x 20.25) into the PPI calculator,
    > is returns 92 ppi. So that seems to be the limit for this camera.


    it's a 4 mp camera, which is much lower than the 16 mp you got from
    your friend's nikon. it's half the resolution (in 2 dimensions). you're
    going in the wrong direction. you want *more* resolution not less.

    > She also sent me a JPG file derived from the CR2 file. It was cropped to
    > a border of about 1" all around. Here are the surprising (to me)
    > properties for that:
    >
    > Compression: JPEG, quality: 100, subsampling OFF
    > Resolution: 300 x 300 DPI
    > Original size: 3156 x 2678 Pixels (8.45 MPixels) (1.18)
    > Current size: 3156 x 2678 Pixels (8.45 MPixels) (1.18)
    > Print size (from DPI): 26.7 x 22.7 cm; 10.5 x 8.9 inches
    > Original colors: 16,7 Millions (24 BitsPerPixel)
    > Current colors: 16,7 Millions (24 BitsPerPixel)
    > Number of unique colors: 177560
    > Disk size: 7.07 MB (7,418,412 Bytes)
    > Current memory size: 24.18 MB (25,355,344 Bytes)
    >
    > This one claims to have 8.45 megapixels. That's more than twice as many
    > as the raw image with about a third of those pixels cropped away. How
    > can that be? Did the JPG processor manufacture (interpolate) new pixels?


    she upscaled it before exporting a jpeg.

    > If I display the two images side by side and scale them so that the
    > portrait is about the same size on my 20 x 13 inch LCD screen, the raw
    > file is a little crisper, but the colors are much more vibrant,
    > especially the reds, oranges, and blues.


    no surprise that the raw file is crisper. it doesn't have interpolated
    pixels.

    as for the vibrancy, that's easy to change. in fact, one of the
    adjustments in photoshop is called vibrancy.

    > The JPG file is not exactly washed out. In fact, if looks very good
    > until I put it up side by side with the raw file.


    the raw file can be adjusted to pretty much anything you want. she
    exported the jpeg with one set of adjustments and you're looking at a
    different set of adjustments.

    > What I found even more interesting was when I put the raw Canon file
    > (4.15 megapixels) side by side next to the raw Nikon file (12.05
    > megapixels and with a little less border. The Canon image is brighter
    > and more vibrant. The Nikon image has a yellow tinge.


    also easy to change. one major advantage of shooting raw is you can
    adjust things with little to no loss on the image.

    > This is particularly noticeable on the frame, which is kind of a bronze
    > color. In the Canon images, the frame is close to the bronze, maybe a
    > little redder. In the Nikon image, it's more brassy, almost golden.
    >
    > The Nikon images were taken indoors in very low light (to minimize
    > reflection) using a long exposure. The Canon images were taken outdoors
    > in the shade (as suggested by Peabody) on a bright afternoon. There is
    > no reflection.


    you could probably match them.

    > In the absense of really good lighting equipment, I think the outdoors
    > in the shade suggestion is the best.


    probably is.
    nospam, Apr 4, 2013
    #2
    1. Advertising

  3. Wolfgang Weisselberg

    Peabody Guest

    Jennifer Murphy says...

    > OK, I think I sorta understand. What matters is the
    > total number of pixels (X x Y). That's the raw data. If
    > I have 3000x3000 pixels, I can have a a 10"x10" print at
    > 300 dpi, or a 20"x20" print at 150 dpi. I use dpi here
    > because I am talking about the dots on the printed page.
    > Right?


    Yes that's right. Printer guys will look at it a bit
    differently, and will sometimes ask you questions like, "Did
    you shoot that at 300 DPI?" Well, you know, that question
    doesn't make much sense, but what he really needs to
    know is whether you have enough pixels to make the print
    size you want and have it look good. The rule of thumb is
    that 300 PPI is a safe harbor that will work for pretty much
    anything. But that doesn't mean a lower resolution
    wouldn't also work. It depends on what's in the picture.
    Some subjects just take lots of resolution to look good.
    But in the overall scheme of things, an oil painting is not
    usually all that demanding - if it's a little soft, you may
    not even notice. And it also depends on what you're
    printing on - a print on canvas is going to be a bit soft
    anyway, but can look wonderful.

    > I'm a little puzzled about the Print Size, because no
    > Resolution was specified. If I plug the pixel count
    > (2496 x 1664) and the print size into my handy little
    > PPI calculator, it cranks out 72 PPI. And if I enter 72
    > in the Resolution fields, the Print size stays the same.


    Digital cameras commonly use 72 DPI, but that's just a
    number - a value in the header. It could just as well be
    419. All that really matters is the pixel resolution.

    For comparison, my 18MP Canon DSLR produces CR2 files of
    33MB in size. So your friend's Canon may be a bit old.
    But I would question the properties shown for the raw file.
    It probably is 8MP, not 4MP. Even so, it looks like she
    upsized to get the jpeg. That usually doesn't work all that
    well because the algorithm has to create pixels.

    > The JPG file is not exactly washed out. In fact, if
    > looks very good until I put it up side by side with the
    > raw file.


    It may be that her software and monitor display raw and jpeg
    files differently than your Irfanview and monitor do. But
    you might see if you can produce a jpeg file that looks as
    good as the raw. Just load it as a CR2 file, then save it
    as a jpeg, and see what happens.

    > The Nikon images were taken indoors in very low light
    > (to minimize reflection) using a long exposure. The
    > Canon images were taken outdoors in the shade (as
    > suggested by Peabody) on a bright afternoon. There is no
    > reflection.


    > In the absense of really good lighting equipment, I
    > think the outdoors in the shade suggestion is the best.


    I'm glad that's working for you, but an overcast day would
    be even better. And I think the difference in lighting
    accounts for the difference in results with the Nikon. You
    have to set the white balance to match the lighting
    situation, but you had so little light to work with indoors,
    it's not surprising it didn't come out all that well.

    I don't know where you are, but if there's a Craigslist city
    near you, you can probably find someone with a high-res
    camera that will do this for you for free. The
    photography for-sale area would be where to post. Or if
    there's a Meetup.com photography group in your area, that would
    also be a good source.

    My artist friend just had some T-shirt and coffee mugs
    printed with images of his paintings, and I photographed
    that stuff - for his website eventually. I don't charge him
    anything because he's a starving artist, and I'm not a pro
    photographer. You can certainly find someone like me where
    you are, but you may have to go outside your circle of
    friends.

    Actually, once you get this done, you could make coffee mugs
    with the painting on them for everyone in the family. :)
    Peabody, Apr 4, 2013
    #3
  4. Jennifer Murphy <> wrote:
    > On Tue, 2 Apr 2013 15:06:18 +0200, Wolfgang Weisselberg
    >>Jennifer Murphy <> wrote:
    >>> On Fri, 29 Mar 2013 19:28:47 -0700, Jennifer Murphy


    >>> Original size: 4256 x 2832 Pixels (12.05 MPixels) (3:2)
    >>> Current size: 4256 x 2832 Pixels (12.05 MPixels) (3:2)


    >>Relevant


    > OK, I think I sorta understand. What matters is the total number of
    > pixels (X x Y). That's the raw data. If I have 3000x3000 pixels, I can
    > have a a 10"x10" print at 300 dpi, or a 20"x20" print at 150 dpi.


    Right (except that it's really ppi, but very many people use
    dpi when the mean ppi and also use dpi when they mean dpi).

    > I use
    > dpi here because I am talking about the dots on the printed page. Right?


    Almost.
    With some technologies ppi equals dpi, where each dot can have
    millions of colours.

    But with most printing methods you need many dots for one pixel.
    An 8 ink inkjet printer has only 8 colours, and probably can't
    merge more than 3 inks in a single dot.[1] (There's a bit more
    freedom if the print head can choose from 2 or 3 amounts of
    ink to squirts.)

    Since a typical "8-bit" pixel can have 16,777,216 combinations
    (8 bit per channel for 256 steps, 3 channels: Red, Green, Blue)
    you need way more colours than a single ink jet dot can have.
    So you need many dots per pixel ... and typical inkjet printers
    do have 2400x9600 or 4800x9600 dpi. Which is way more than
    300x300 ppi (and you'd need 9 or 18 GIGApixel for 20"x20").
    With these dpi you have 256 dots per pixel assuming 300x300 ppi.
    Which means you can just get about 16.7 million different
    colours with cyan, magenta and yellow.

    Unfortunately, with the typical 8-bit pixel in, say, a JPEG
    the colour steps aren't linear, but gamma encoded, which
    means that the steps are visually similar sized. Thus the
    driver may need to dither over more than the area of one pixel.


    Worse, some colours can't be represented by the printer,
    yet other colours are outside the sRGB colour space. So you
    don't have the full colour range no matter what you do, nor
    can you trivially use the full colour range of the printer.

    You probably want the printer and the monitor to match, and
    both to match other people's monitors. You also probably
    want your camera to record the colours faithfully (though
    you certainly can devitate from "faithful to the real world"
    for artistic reasons).
    Thus the raise of colour management (which is a whole science,
    but the basic idea it to *measure* colour patches[2] and build
    a mathematical model (using software) from the "colour values
    XYZ => gives result ABC" data to a "wanted result abc =>
    you need colour values ijk" and use software which translate
    between a colourspace and the next (e.g. between the standard
    colourspace sRGB and the monitor's specific colourspace).
    The idea is that, e.g. a print --- given normlight --- can be
    held next to your monitor and look as identical as possible
    (given the limitations of both). Or a print --- given the
    pre-set light --- can be held to the painting it's from and
    it's as close as it gets.


    When almost all dot positions are white (i.e. with light
    colours) the few coloured dots look bad visually. Hence light
    cyan, light magenta, grey, where you can use more coloured
    dots without getting darker.


    > I snipped the rest because I'm not going to take multiple shots and
    > stitch them together (horrors!). I think you were joking with me anyway.


    Nope. Google >gigapixel panorama<.
    You don't have to do that, of course, a couple photos done
    right work wonders.

    The alternative is of course an expensive medium format
    equipment or large format scanback unit.

    Or less than 300 ppi.

    > And I'm not going to go get a camera that I probably can't lift


    Medium format you can lift easily. Large format ... well ....

    > so I can
    > get a raw image at 300 ppi for a 20x24 inch image. As you and others
    > have said, I probably already have something that will be good enough.


    OK.


    > However, I still have friends with a photography habit who are willing
    > to play around on this project. And I am learning a few things.


    > Another friend came over with a Canon DSLR. I forgot (again) to get the
    > model number, but it was huge. She just sent me two images.


    huge?

    > One is the raw camera file. It has a .CR2 extension. Here are the
    > properties:


    > Original size: 2496 x 1664 Pixels (4.15 MPixels) (3:2)


    pretty small. Could be an original 1D from 2001. That's a
    rather huge camera.


    > I'm a little puzzled about the Print Size, because no Resolution was
    > specified. If I plug the pixel count (2496 x 1664) and the print size
    > into my handy little PPI calculator, it cranks out 72 PPI. And if I
    > enter 72 in the Resolution fields, the Print size stays the same.


    A default value.

    > Original size: 3156 x 2678 Pixels (8.45 MPixels) (1.18)


    Then it's not a 1D, but a later one, unless it's upsampled.


    > This one claims to have 8.45 megapixels. That's more than twice as many
    > as the raw image with about a third of those pixels cropped away. How
    > can that be? Did the JPG processor manufacture (interpolate) new pixels?


    Could be. Some cameras can also down-scale the CR2.


    > If I display the two images side by side and scale them so that the
    > portrait is about the same size on my 20 x 13 inch LCD screen, the raw
    > file is a little crisper, but the colors are much more vibrant,
    > especially the reds, oranges, and blues.


    > The JPG file is not exactly washed out. In fact, if looks very good
    > until I put it up side by side with the raw file.


    The RAW is open to interpretation --- and it *needs* to be
    interpreted. Your software must interpret it (or use a preview
    JPEG). You're probably falling for 'stronger colours == better'
    (same with "sharpened more == better, even when the sharpening
    artefacts are already well noticeable).

    The real test is if the print does match the painting!
    If the print is much more vibrant than the painting, what good
    is it?


    > What I found even more interesting was when I put the raw Canon file
    > (4.15 megapixels) side by side next to the raw Nikon file (12.05
    > megapixels and with a little less border. The Canon image is brighter
    > and more vibrant. The Nikon image has a yellow tinge.


    As I said, you're re-interpreting the RAW. You need to
    interprete RAW files from each camera differently --- they
    use different colour filters (the 'red' from Canon differs
    from the 'red' from Nikon, etc.)


    > This is particularly noticeable on the frame, which is kind of a bronze
    > color. In the Canon images, the frame is close to the bronze, maybe a
    > little redder. In the Nikon image, it's more brassy, almost golden.


    That's why you want to photograph a colour chart, so you can
    correct the specifics of the camera and the light situation
    --- so that a specific red stays that specific red.


    > The Nikon images were taken indoors in very low light (to minimize
    > reflection) using a long exposure.


    This doesn't help. You need to expose longer by the same
    amount as you dim the light --- which means the same amount
    of light collected over the exposure time

    You need to change the direction of the light (e.g. change
    the angle, get the light to come from more directions so that
    the single direction for one *specific* reflection has a much
    smaller percentage of light coming in which it can reflect, ...)


    Of course, using say white bed sheets to spread the light
    without increasing the power of the light causes less light
    to reach the painting. :)

    > The Canon images were taken outdoors
    > in the shade (as suggested by Peabody) on a bright afternoon. There is
    > no reflection.


    There is a marked difference between "in the shade" and "in
    the sun", between "sunny" and "overcast", never mind between
    sunlight and artificial light![3]

    Thus you need to correct the image differently even if you use
    the identical camera and camera settings. Camera settings
    don't affect the RAW file data, only the RAW file metadata,
    which may or may not be used as a default value for interpreting
    the RAW data. (Of course, the camera maker's RAW converter
    can read and use the metadata and thus the camera settings,
    and JPEGs must use the camera settings. So camera settings
    can be important.)


    > In the absense of really good lighting equipment, I think the outdoors
    > in the shade suggestion is the best.


    This is a very blue-ish light (all the blue from the sky!)
    and also depends on what areas do reflect light towards the
    painting --- large grass areas or under trees cause more green,
    for example.

    Overcast is probably the most neutral (the same light from all
    directions), it's also very flat (coming from all directions,
    it's a really huge lightbox), which can help with
    reflections.


    -Wolfgang

    [1] Too much ink and the paper won't absorb it --- the ink
    will run, the paper won't stay flat ...
    [2] on your monitor, using a colourimeter or spectrometer
    (and repeat that every few weeks, as monitors do age),
    on your printer (there are service providers that you can
    sent prints of patches to to be measured (you need one per
    printer-ink-paper combination), and there are also generic
    profiles for printertype-selected inks-selected papers),
    on your camera (photographing normed patches, e.g colour
    checker, in a given light situation (with some dependence
    on ISO, lens, aperture,))
    [3] Incandescent light sources are continuous, that is they
    don't leave out frequencies --- but cooler (redder) ones
    are very different from hotter (bluer) ones.
    Flourescent lights often have only 3 frequency bands,
    better ones have 5 ... some colours look completely
    different under them. LED "white" lights are also not
    continuous, they're often blue light + a yellow phosphor
    (which age differently from each other).
    Wolfgang Weisselberg, Apr 5, 2013
    #4
  5. Eric Stevens <> wrote:
    > On Fri, 5 Apr 2013 20:36:28 +0200, Wolfgang Weisselberg


    > [An aside for Wolfgang ... ]


    >>> This is particularly noticeable on the frame, which is kind of a bronze
    >>> color. In the Canon images, the frame is close to the bronze, maybe a
    >>> little redder. In the Nikon image, it's more brassy, almost golden.


    >>That's why you want to photograph a colour chart, so you can
    >>correct the specifics of the camera and the light situation
    >>--- so that a specific red stays that specific red.


    > ... and that's why DxO Pro has the ability to nominate the camera as
    > part of its colour correction setup.


    Every single raw converter that at least semi-works has to
    have a fairly good idea of the filter spectra of a given camera.

    Unfortunately a "colour correction" without measuring the
    light spectrum on the subject at least roughly (e.g. with a
    colour chart on the subject) is a nice gag --- not something
    other raw converters don't do.


    > And they really are different as
    > Jennifere described.


    Which might simply be a different automatic white balance from
    the two cameras.

    -Wolfgang
    Wolfgang Weisselberg, Apr 10, 2013
    #5
  6. Eric Stevens <> wrote:
    > On Wed, 10 Apr 2013 21:23:43 +0200, Wolfgang Weisselberg
    >>Eric Stevens <> wrote:


    >>> And they really are different as
    >>> Jennifere described.


    >>Which might simply be a different automatic white balance from
    >>the two cameras.


    > I find that in DxO Pro there is a visible change in the image if I
    > tell the software I am using a Canon something or other instead of my
    > D300.


    Obvious and irrelevant.

    Obvious: a different maker and a different camera means
    different RGB filter spectra.

    Irrelevant: The problem was:
    2 photos, made by 2 different cameras, in 2 very different
    light situations
    and not:
    1 camera, one light situation, but once you lie about the
    camera and once not


    > Nothing else has been changed. The ICC profile remains the same.


    Which ICC profile? That of the camera? Or that of your
    monitor?

    > So too does the colour temperature and the spectrum at the time the
    > image was taken. But DxO says that even if all these things remain
    > constant, changing the camera body will change the apparent colours of
    > the image.


    The idea is that IF you match the bodies you used to the bodies
    you claim to have used THEN the differences in the RGB filter
    spectra are compensated by the different treatment of the camera
    RGB values according to the bodies you claim to have used.

    DxO even claims some magical camera calibration --- what
    else can it mean but that such camera filter differences are
    corrected to one (assumed) correct representation?

    -Wolfgang
    Wolfgang Weisselberg, Apr 12, 2013
    #6
  7. Eric Stevens <> wrote:
    > On Sat, 13 Apr 2013 00:41:28 +0200, Wolfgang Weisselberg


    >>The idea is that IF you match the bodies you used to the bodies
    >>you claim to have used THEN the differences in the RGB filter
    >>spectra are compensated by the different treatment of the camera
    >>RGB values according to the bodies you claim to have used.


    >>DxO even claims some magical camera calibration --- what
    >>else can it mean but that such camera filter differences are
    >>corrected to one (assumed) correct representation?


    > Now you've got it.


    > It's the DxO standard specification and you can make your edits from
    > there.


    And it's what every other converter has to do as well.

    It's like me claiming a magical oxygen extraction and
    transformation in my body. Turns oxygen into carbon dioxide.
    Reliably and constantly. Completely magical. No other body
    does that!

    -Wolfgang
    Wolfgang Weisselberg, Apr 15, 2013
    #7
  8. Eric Stevens <> wrote:
    > On Mon, 15 Apr 2013 18:52:23 +0200, Wolfgang Weisselberg
    >>Eric Stevens <> wrote:
    >>> On Sat, 13 Apr 2013 00:41:28 +0200, Wolfgang Weisselberg


    >>>>The idea is that IF you match the bodies you used to the bodies
    >>>>you claim to have used THEN the differences in the RGB filter
    >>>>spectra are compensated by the different treatment of the camera
    >>>>RGB values according to the bodies you claim to have used.


    >>>>DxO even claims some magical camera calibration --- what
    >>>>else can it mean but that such camera filter differences are
    >>>>corrected to one (assumed) correct representation?


    >>> Now you've got it.


    >>> It's the DxO standard specification and you can make your edits from
    >>> there.


    >>And it's what every other converter has to do as well.


    > The difference is that only DxO (as far as I know) has


    .... a marketing department that claims that that part, which is as
    necessary and as unsexy as working brakes on a car, into
    something magical-mystical superior?

    > a procedure for
    > precisely measuring the colour sensitivity of individual types of
    > camera bodies.


    How comes you think that?

    What specialized materiel and procedures does it take to
    measure the transmission of a filter system when one already
    has a electron counting sensor behind the filter system?

    You DO know even you can photograph a number of well defined
    patches under well defined light conditions and evaluate the
    results? You don't even need a light source tuned/filtered
    to specific narrow banded frequencies, e.g. with a grating or
    a prism?

    >>It's like me claiming a magical oxygen extraction and
    >>transformation in my body. Turns oxygen into carbon dioxide.
    >>Reliably and constantly. Completely magical. No other body
    >>does that!


    > That's a specious comparison. I'm sure you know that.


    Do you think so? Why is it specious when I claim something
    as natural as breathing is something unique and special, but
    when DxO claims the equivalent, you fall for it hook, line,
    sinker and fishing rod?

    -Wolfgang
    Wolfgang Weisselberg, May 2, 2013
    #8
  9. Eric Stevens <> wrote:
    > On Thu, 2 May 2013 23:21:45 +0200, Wolfgang Weisselberg
    >>Eric Stevens <> wrote:
    >>> On Mon, 15 Apr 2013 18:52:23 +0200, Wolfgang Weisselberg
    >>>>Eric Stevens <> wrote:
    >>>>> On Sat, 13 Apr 2013 00:41:28 +0200, Wolfgang Weisselberg


    >>>>>>The idea is that IF you match the bodies you used to the bodies
    >>>>>>you claim to have used THEN the differences in the RGB filter
    >>>>>>spectra are compensated by the different treatment of the camera
    >>>>>>RGB values according to the bodies you claim to have used.


    >>>>>>DxO even claims some magical camera calibration --- what
    >>>>>>else can it mean but that such camera filter differences are
    >>>>>>corrected to one (assumed) correct representation?


    >>>>> Now you've got it.


    >>>>> It's the DxO standard specification and you can make your edits from
    >>>>> there.


    >>>>And it's what every other converter has to do as well.


    >>> The difference is that only DxO (as far as I know) has


    >>... a marketing department that claims that that part, which is as
    >>necessary and as unsexy as working brakes on a car, into
    >>something magical-mystical superior?


    >>> a procedure for
    >>> precisely measuring the colour sensitivity of individual types of
    >>> camera bodies.


    >>How comes you think that?


    >>What specialized materiel and procedures does it take to
    >>measure the transmission of a filter system when one already
    >>has a electron counting sensor behind the filter system?


    >>You DO know even you can photograph a number of well defined
    >>patches under well defined light conditions and evaluate the
    >>results? You don't even need a light source tuned/filtered
    >>to specific narrow banded frequencies, e.g. with a grating or
    >>a prism?


    > You are rambling and I am sure you know that.


    Then you would accept a short "You don't have a clue.
    They don't do anything special except for marketing for which
    you fell."?

    > If you know of any other
    > software which specifically tests for colour sensitivity under a range
    > of different circumstances and then applies it to image correction, I
    > am sure you would have told us.


    Let's see: DPP. AfterShot Pro. Lightroom. ...
    In fact any software that offers WB presets (different
    circumstances) and manages to produce OK colours.

    >>>>It's like me claiming a magical oxygen extraction and
    >>>>transformation in my body. Turns oxygen into carbon dioxide.
    >>>>Reliably and constantly. Completely magical. No other body
    >>>>does that!


    >>> That's a specious comparison. I'm sure you know that.


    >>Do you think so? Why is it specious when I claim something
    >>as natural as breathing is something unique and special, but
    >>when DxO claims the equivalent, you fall for it hook, line,
    >>sinker and fishing rod?


    > Because its not as natural as breathing and (as far as I know) DxO are
    > the only people who do it outside of a lab.


    I don't know that you are a human, therfore you must be a dog.
    That's by your logic, not by mine.

    -Wolfgang
    Wolfgang Weisselberg, May 7, 2013
    #9
  10. Eric Stevens <> wrote:
    > On Tue, 7 May 2013 22:59:03 +0200, Wolfgang Weisselberg
    >>Eric Stevens <> wrote:
    >>> On Thu, 2 May 2013 23:21:45 +0200, Wolfgang Weisselberg
    >>>>Eric Stevens <> wrote:
    >>>>> On Mon, 15 Apr 2013 18:52:23 +0200, Wolfgang Weisselberg
    >>>>>>Eric Stevens <> wrote:
    >>>>>>> On Sat, 13 Apr 2013 00:41:28 +0200, Wolfgang Weisselberg


    >>>>You DO know even you can photograph a number of well defined
    >>>>patches under well defined light conditions and evaluate the
    >>>>results? You don't even need a light source tuned/filtered
    >>>>to specific narrow banded frequencies, e.g. with a grating or
    >>>>a prism?


    >>> You are rambling and I am sure you know that.


    >>Then you would accept a short "You don't have a clue.
    >>They don't do anything special except for marketing for which
    >>you fell."?


    > No, I wouldn't accept that. I would accept that you are trying to be
    > difficult in defence of an indefensible position.


    I would accept that you are misled by (arguably clever)
    marketing from DxO, or paid by DxO.


    >>> If you know of any other
    >>> software which specifically tests for colour sensitivity under a range
    >>> of different circumstances and then applies it to image correction, I
    >>> am sure you would have told us.


    >>Let's see: DPP. AfterShot Pro. Lightroom. ...
    >>In fact any software that offers WB presets (different
    >>circumstances) and manages to produce OK colours.


    > Please tell us how and where these various products go about testing
    > the colour sensitivities of various cameras and how the results are
    > embedded in their software.


    Please tell us *how* and *where* DxO goes about to testing the
    colour sensitivities of various cameras and how the results
    are embedded in their software.

    As to the others: They make test shots and embed the results
    of the results of them in the conversion from RAW file data to
    a proper TIFF or JPEG or similar with the correct colour space.


    > It will not be sufficient for you to explain how these products
    > provide knobs which the operator can twiddle until they obtain a
    > result with which they are subjectively satisfied.


    It will not be sufficient for you to recite advertizing
    claims, I want details from you.


    >>>>>>It's like me claiming a magical oxygen extraction and
    >>>>>>transformation in my body. Turns oxygen into carbon dioxide.
    >>>>>>Reliably and constantly. Completely magical. No other body
    >>>>>>does that!


    >>>>> That's a specious comparison. I'm sure you know that.


    >>>>Do you think so? Why is it specious when I claim something
    >>>>as natural as breathing is something unique and special, but
    >>>>when DxO claims the equivalent, you fall for it hook, line,
    >>>>sinker and fishing rod?


    >>> Because its not as natural as breathing and (as far as I know) DxO are
    >>> the only people who do it outside of a lab.


    >>I don't know that you are a human, therfore you must be a dog.
    >>That's by your logic, not by mine.


    Please bark now!

    -Wolfgang
    Wolfgang Weisselberg, May 15, 2013
    #10
  11. Wolfgang Weisselberg

    PeterN Guest

    On 5/17/2013 12:10 AM, Eric Stevens wrote:
    > On Wed, 15 May 2013 20:07:58 +0200, Wolfgang Weisselberg
    > <> wrote:
    >
    >> Eric Stevens <> wrote:
    >>> On Tue, 7 May 2013 22:59:03 +0200, Wolfgang Weisselberg
    >>>> Eric Stevens <> wrote:
    >>>>> On Thu, 2 May 2013 23:21:45 +0200, Wolfgang Weisselberg
    >>>>>> Eric Stevens <> wrote:
    >>>>>>> On Mon, 15 Apr 2013 18:52:23 +0200, Wolfgang Weisselberg
    >>>>>>>> Eric Stevens <> wrote:
    >>>>>>>>> On Sat, 13 Apr 2013 00:41:28 +0200, Wolfgang Weisselberg

    >>
    >>>>>> You DO know even you can photograph a number of well defined
    >>>>>> patches under well defined light conditions and evaluate the
    >>>>>> results? You don't even need a light source tuned/filtered
    >>>>>> to specific narrow banded frequencies, e.g. with a grating or
    >>>>>> a prism?

    >>
    >>>>> You are rambling and I am sure you know that.

    >>
    >>>> Then you would accept a short "You don't have a clue.
    >>>> They don't do anything special except for marketing for which
    >>>> you fell."?

    >>
    >>> No, I wouldn't accept that. I would accept that you are trying to be
    >>> difficult in defence of an indefensible position.

    >>
    >> I would accept that you are misled by (arguably clever)
    >> marketing from DxO, or paid by DxO.
    >>
    >>
    >>>>> If you know of any other
    >>>>> software which specifically tests for colour sensitivity under a range
    >>>>> of different circumstances and then applies it to image correction, I
    >>>>> am sure you would have told us.

    >>
    >>>> Let's see: DPP. AfterShot Pro. Lightroom. ...
    >>>> In fact any software that offers WB presets (different
    >>>> circumstances) and manages to produce OK colours.

    >>
    >>> Please tell us how and where these various products go about testing
    >>> the colour sensitivities of various cameras and how the results are
    >>> embedded in their software.

    >>
    >> Please tell us *how* and *where* DxO goes about to testing the
    >> colour sensitivities of various cameras and how the results
    >> are embedded in their software.
    >>
    >> As to the others: They make test shots and embed the results
    >> of the results of them in the conversion from RAW file data to
    >> a proper TIFF or JPEG or similar with the correct colour space.

    >
    > So they make test shots. With how cameras and of what type?
    >
    > Assuming that is what they do, how does the software know which camera
    > it is supposed to be emulating?
    >>
    >>
    >>> It will not be sufficient for you to explain how these products
    >>> provide knobs which the operator can twiddle until they obtain a
    >>> result with which they are subjectively satisfied.

    >>
    >> It will not be sufficient for you to recite advertizing
    >> claims, I want details from you.

    >
    > I'm not sure that you will read them. I don't think you have read
    > those I have given you already. However, try these:
    >
    > Go to http://dxo.com/intl/photo/dxo_optics_pro/features and scroll
    > down to:
    >
    > "Visit our lab
    >
    > Come take an exclusive tour and check out the different calibration
    > process steps for cameras and lenses in our laboratory."
    >
    > Click on the adjacent video and watch:
    >
    > That DxO really does seem to believe they have the ability to adjust
    > colour renditions to match different cameras is made clear in:
    >
    > http://www.dxo.com/en/photo/dxo_optics_pro/features/raw_conversion
    > "Each manufacturer, and often each camera model, is characterized by
    > how it renders color. But what color range do you want for your
    > images? DxO Labs has tested a wide range of digital cameras, giving
    > you the choice of applying their different color renderings to your
    > photos, regardless of the camera that you actually use."
    >


    Saw a live demo last night of the full DXO line. We put the presenter to
    the proof, and he came through with flying colors. It was a two hour
    demonstration, of which almost one hour was reserved for our skeptical
    questioning.
    DXO Viewpoint did perspective control with greater ease, and in less
    time than the demo of that feature in Creative Cloud.

    --
    PeterN
    PeterN, May 17, 2013
    #11
  12. Wolfgang Weisselberg

    PeterN Guest

    On 5/17/2013 6:01 PM, Eric Stevens wrote:
    > On Fri, 17 May 2013 11:34:28 -0400, PeterN
    > <> wrote:


    <snip>

    >> Saw a live demo last night of the full DXO line. We put the presenter to
    >> the proof, and he came through with flying colors. It was a two hour
    >> demonstration, of which almost one hour was reserved for our skeptical
    >> questioning.
    >> DXO Viewpoint did perspective control with greater ease, and in less
    >> time than the demo of that feature in Creative Cloud.

    >
    > It's not perfect but it seems to do things that no other application
    > that I know of does.
    >
    > I was particularly impressed with what even the old software can do
    > with a fisheye lens.
    > http://www.kenrockwell.com/tech/dxo/optics-pro.htm
    > Not that I have got a fisheye lens.
    >


    No it's not. I was particularly impressed with it's noise reduction and
    highlight recovery capabilities. Some stores here have a special
    involving a significant price reduction. The elite verson is $200 in
    stead of $300, with a more significant reduction in Viewpoint, and
    filmpack. I see little use for
    filmpack, but will be ordering Viewpoint.

    --
    PeterN
    PeterN, May 18, 2013
    #12
  13. Eric Stevens <> wrote:
    > On Wed, 15 May 2013 20:07:58 +0200, Wolfgang Weisselberg
    >>Eric Stevens <> wrote:
    >>> On Tue, 7 May 2013 22:59:03 +0200, Wolfgang Weisselberg
    >>>>Eric Stevens <> wrote:
    >>>>> On Thu, 2 May 2013 23:21:45 +0200, Wolfgang Weisselberg
    >>>>>>Eric Stevens <> wrote:
    >>>>>>> On Mon, 15 Apr 2013 18:52:23 +0200, Wolfgang Weisselberg
    >>>>>>>>Eric Stevens <> wrote:
    >>>>>>>>> On Sat, 13 Apr 2013 00:41:28 +0200, Wolfgang Weisselberg


    >>>>>>You DO know even you can photograph a number of well defined
    >>>>>>patches under well defined light conditions and evaluate the
    >>>>>>results? You don't even need a light source tuned/filtered
    >>>>>>to specific narrow banded frequencies, e.g. with a grating or
    >>>>>>a prism?


    >>>>> You are rambling and I am sure you know that.


    >>>>Then you would accept a short "You don't have a clue.
    >>>>They don't do anything special except for marketing for which
    >>>>you fell."?


    >>> No, I wouldn't accept that. I would accept that you are trying to be
    >>> difficult in defence of an indefensible position.


    >>I would accept that you are misled by (arguably clever)
    >>marketing from DxO, or paid by DxO.


    >>>>> If you know of any other
    >>>>> software which specifically tests for colour sensitivity under a range
    >>>>> of different circumstances and then applies it to image correction, I
    >>>>> am sure you would have told us.


    >>>>Let's see: DPP. AfterShot Pro. Lightroom. ...
    >>>>In fact any software that offers WB presets (different
    >>>>circumstances) and manages to produce OK colours.


    >>> Please tell us how and where these various products go about testing
    >>> the colour sensitivities of various cameras and how the results are
    >>> embedded in their software.


    >>Please tell us *how* and *where* DxO goes about to testing the
    >>colour sensitivities of various cameras and how the results
    >>are embedded in their software.


    >>As to the others: They make test shots and embed the results
    >>of the results of them in the conversion from RAW file data to
    >>a proper TIFF or JPEG or similar with the correct colour space.


    > So they make test shots. With how cameras and of what type?


    well, since you're the one asking, they probably use a
    point'n'shoot from 2004 and from a completely different
    manufacturer to find out what colours give what response
    in a 2012/2013 DSLR. For added fun, they roll dice as to
    what lighting conditions should be used (and don't note that
    choice down). Lastly, I guess they let an ape mix and apply the
    paint from different paint suppliers (with different pigments
    and dye types) each time.

    Only DxO has understood that you probably need to shoot the
    test scenarios with the camera you're going to measure the
    colour response for and that it might be a good idea to have
    a well known lighting situation: even and of known spectral
    composition, and that you might as well go ahead and use
    colour targets of which the exact colour response is well
    known and measured.

    Then, of course, only DxO has figured out what colour spectrum
    will actually be reflected when one has a well defined,
    well behaved target (no flourescent whiteners, for example)
    and a well defined light shining onto it. They probably have
    patents on adding the numbers involved, too, not that anyone
    else could come up with it. Then DxO does something
    completely revolutionary that noone else in the world does:
    Use RAW! They actually use RAW to find out how strong the
    response is to each of that patches, given that they know how
    much light entered the camera and how the response varies
    from patch to patch of the same exposure.

    Then DxO uses their proprietary, secret, patented knowledge
    to compute the transmission spectrum of each of the R, G, G,
    and B filters (yep, 2 different G filters in some cameras)
    and use *that* to calculate that if the light source is of
    spectrum X then the colour Y will be recorded as a
    (A,B,C,D)-RGGB-quadrupel.

    Maybe they even check that by tuning their light source to a
    given spectrum, run their math and check if the patches come
    out as calculated.


    > Assuming that is what they do, how does the software know which camera
    > it is supposed to be emulating?


    Well, of course they can't know, because only DxO has the secret
    sauce necessary to read the metadata in the RAW. You know,
    where such stuff as exposure time or focal length or aperture
    is also noted down.


    >>> It will not be sufficient for you to explain how these products
    >>> provide knobs which the operator can twiddle until they obtain a
    >>> result with which they are subjectively satisfied.


    >>It will not be sufficient for you to recite advertizing
    >>claims, I want details from you.


    > I'm not sure that you will read them. I don't think you have read
    > those I have given you already. However, try these:


    > Go to http://dxo.com/intl/photo/dxo_optics_pro/features and scroll
    > down to:


    > "Visit our lab


    > Come take an exclusive tour and check out the different calibration
    > process steps for cameras and lenses in our laboratory."


    > Click on the adjacent video and watch:


    As to colour correction, all they say is that they shoot a
    colour target (with only 24 patches!) under differently
    filtered lights. This is neither complicated nor unique and
    way less powerful than what I wrote above.


    > That DxO really does seem to believe they have the ability to adjust
    > colour renditions to match different cameras is made clear in:


    > http://www.dxo.com/en/photo/dxo_optics_pro/features/raw_conversion
    > "Each manufacturer, and often each camera model, is characterized by
    > how it renders color. But what color range do you want for your
    > images? DxO Labs has tested a wide range of digital cameras, giving
    > you the choice of applying their different color renderings to your
    > photos, regardless of the camera that you actually use."


    You mean same as I do when I shoot a colour target, e.g. a
    QPcard 201 under a given light and use some software or plugin
    to correct the differences from the *known* patches of the
    card to the *measured* RGB response of the sensor?

    Except they don't do it under the light I am using, but only
    under some presets, like "sunny", "open shade", "tungsten"
    and "flourescent"? Hello --- how do you think these presets
    work anyhow? How do you think the RGB values, which are
    different from each camera to each other camera --- because
    the filters are changed --- can be converted to reasonable
    colour images otherwise?


    > http://1.static.img-dpreview.com/files/articles/6432558797/250/protect_autooff.png?v=2169
    > Shows the screen module by means of which the user can select the
    > camera whose response the user wishes to emulate.


    So basically they allow 'cross processing' with a menu
    selection.

    Well, in AfterShot Pro I can feed it any random icc or icm as
    the camera profile --- and yes, of course does the image
    change visibily.

    As they say:
    | Input profiles
    |
    | Input profiles are used as the starting point for color
    | management. Every image file in Corel AfterShot Pro has an
    | input profile. For supported RAW files, Corel performs
    | detailed color calibration and profiling techniques to
    | produce unique Color Profiles for the various supported
    | cameras. [...]
    http://product.corel.com/help/AfterShot/540111115/Main/EN/Doc/index.html

    Well, it seems they do what DxO does regarding to colour, they
    just don't produce the same amount of drama and self-adulation
    about it.

    See also here:
    http://product.corel.com/help/AfterShot/540111115/Main/EN/Doc/index.html?color_tools.html
    (scroll to bottom, "Color Management")

    -Wolfgang
    Wolfgang Weisselberg, May 18, 2013
    #13
    1. Advertising

Want to reply to this thread or ask your own question?

It takes just 2 minutes to sign up (and it's free!). Just click the sign up button to choose a username and then you can ask your own questions on the forum.
Similar Threads
  1. nospam

    Re: How to photograph an oil painting

    nospam, Mar 30, 2013, in forum: Digital Photography
    Replies:
    19
    Views:
    269
    PeterN
    Apr 1, 2013
  2. RichA

    Re: How to photograph an oil painting

    RichA, Mar 30, 2013, in forum: Digital Photography
    Replies:
    36
    Views:
    524
    James Silverton
    Apr 5, 2013
  3. nick c

    Re: How to photograph an oil painting

    nick c, Mar 30, 2013, in forum: Digital Photography
    Replies:
    0
    Views:
    211
    nick c
    Mar 30, 2013
  4. Peabody

    Re: How to photograph an oil painting

    Peabody, Apr 1, 2013, in forum: Digital Photography
    Replies:
    2
    Views:
    183
    Wolfgang Weisselberg
    Apr 1, 2013
  5. nospam

    Re: How to photograph an oil painting

    nospam, Apr 2, 2013, in forum: Digital Photography
    Replies:
    6
    Views:
    203
    Peabody
    Apr 3, 2013
Loading...

Share This Page