Search the CONFOCAL archive at
http://listserv.acsu.buffalo.edu/cgi-bin/wa?S1=confocal
Dear Jerry,
This can of worms squiggles most strongly exactly do to the reactive nature of the perception of image
processing. Photoshop is not the problem. Not the solution, either.
Image processing should be approached as an act of translating the signal detected and produce by an
instrument that responds to intensities into a representation that can be recognized by the human
visual system designed to detect edges, changes and contrast. Image process is all too often viewed as
random actions to address personal esthetics. The entire process of recording a microscopic image is a
series of translations, beginning with translating the sample into a diffraction pattern at the back of the
objective lens, then into electrons, into a series of digital values. We spend huge amounts of money,
time and procedures to ensure that these translations are carried out with minimal losses at each stage
of the instrumentation. Then most people screw it up at the very end by using arbitrary actions in
image processing software to make something look "good" (meeting pre-conceived notions).
Information losses and histogram changes occur at every stage, most affect gamma. Even the choice of
excitation/emission parameters (filter, laser lines, etc) will skew the gamma. As you point out, the
resulting image must be processed in some manner to both compensate for instrument effects as to
render it meaningful to our visual system.
Regards,
Glen
Glen MacDonald
Core for Communication Research
Virginia Merrill Bloedel Hearing Research Center
Box 357923
University of Washington
Seattle, WA 98195-7923
(206) 616-4156
On Sat, 16 Jun 2007, Jerry Sedgewick wrote:
> Search the CONFOCAL archive at
> http://listserv.acsu.buffalo.edu/cgi-bin/wa?S1=confocal
>
> Hello All,
>
> The can of worms opened the wrong way. I wasn't clear enough about the point of
> the last message. Here is the point:
>
> So far, what has been carefully laid out by Nature and other journals, and by
> Doug's excellent contributions at the University of Arizona's webpage, are the
> Do's and Don'ts of Photoshop manipulation. I call this a REACTIVE response to
> the use of Photoshop.
>
> What hasn't been carefully thought out and made available are the Photoshop
> manipulations that MUST be done to obtain reproduction free of the very thing
> the Do's and Don'ts address: misrepresentation of visual data. I call this a
> PROACTIVE response.
>
> The proactive response posits that most every image must be manipulated in
> Photoshop in order to get reproduction that is NOT a misrepresentation of data:
> Photoshop HAS to be used. It also assumes that the output devices/printing
> presses will NOT, in most circumstances, create an adequate representation of
> images without the educated use of Photoshop to conform color/grayscale values
> to the output device.
>
> For example, the offset/Black Level settings chosen when properly acquiring
> images on a confocal result in the darkest gray value at or close to zero. To
> get reproduction sans misrepresentation on a printing press (and virtually
> every other output device) requires that the darkest black be set at or greater
> than a grayscale value of 20 on an 8-bit scale. That was suggested by Michael
> when describing how photographs were printed in the darkroom: no pure blacks.
>
> By the way, the best means for retaining color and contrast values when
> changing the mode from RGB to CMYK, especially for darkfield images, lies in
> setting the min black value for the RGB channels at 20 or a K: value of 90% (or
> less). Don't rely upon what the monitor shows, but upon the numbers.
>
> Another example: The pure greens used in RGB color space simply do not
> translate to CMYK space. That's why so many published greens look saturated,
> dull and contrast-less. If a scientist is allowed to submit an RGB image for
> publication (versus a CMYK image), they are left to rely upon the printing
> press to make the conversion. Nine times out of ten in my own experience, the
> printing press produces a misrepresentation of the data. That's because
> scientific images are a world unto themselves, and most the rest of the world
> cannot work with these images. Once I submitted a green-colorized neuron to the
> "color theory" listserve, comprised of color correction professionals mostly
> based in North America. Several tried to make the conversion while retaining
> contrast values, but not a one could produce an adequate representation of the
> color green while keeping the relationship of contrast values.
>
> By the way, the conversion CAN be done without using Pantone colors. If color
> values of green are altered to include more yellow or more blue, a far better
> representation of the image can be had when it's published. Start with the
> grayscale image (or convert from RGB using Channel Mixer in Photoshop: do NOT
> change mode from RGB to grayscale!!!). Then convert from grayscale to RGB (what
> seems counterintuitive), and then adjust channels one by one. Using the output
> slider in Levels, adjust the red channel (from dropdown list) to 160, the green
> to 255 (don't adjust), and the blue to 140 for a yellower green. A bluish-green
> can be had by adjusting red to 110, green 255 (don't adjust), blue 160.
>
> Again, a monitor cannot be used to evaluate: trust the numbers and the results
> when transferring from RGB to CMYK (If the black level of a CRT computer screen
> is set properly and ambient lights are off, then colors are better evaluated).
> Also, these are not hard and fast numbers, as variation from image to image
> exists.
>
> Granted, this method does not retain the pure green that may be necessary to
> show such things as coexistence/colocalization, but published visualizations
> seem like a lot of fiction, anyway, when scatterplots do not accompany the
> visual data. Too often values must be arbitrarily saturated in order to show
> yellow when red and green are merged. Furthermore, colorization itself is
> somewhat arbitrary.
>
> Adobe itself suggests methods that do not fit the world of science. For
> example, their recommendation is that 16-bit to 8-bit images be converted using
> mode change. With many manufacturers of scientific cameras and accompanying
> software, 16-bit images--though read in newer versions of Photoshop--are
> truncated to 8-bit (ostensibly because of inabilities to read these files. I
> haven't checked CS3: maybe that's been changed). A mode change to 8-bit results
> in a loss of grayscale values (gone! poof!). These images, though read in
> Photoshop, must be adjusted in Levels first, and then the mode is changed to
> 8-bit. Note that this method doesn't apply for ALL images: truncation of values
> can be seen on the histogram in Photoshop to verify that the grayscale values
> have been truncated. I've had this problem with TIF files from particular
> manufacturers, who's camera systems are used at 3 workstations.
>
> I worked for almost 10 years as a commercial photographer in the former world
> of film, and also spent time in the darkroom in the 80's working for a single
> department (when these had money), before I entered the world of digital
> imaging and light microscopy/core lab. It pains me to see so many confounded by
> the process of RGB to CMYK conversion, along with the use of incorrect methods
> for conforming images to output (or not).
>
> Publications do not send methods to scientists for adjusting color and contrast
> values which is a shame (or a sham), because these images are too often
> misrepresentations when published. All the author gets are maddening
> requirements to keep output pixel resolutions high while simultaneously keeping
> file sizes low for web transfer.
>
> My hope was to get a feel for how others view the idea of "misrepresentation
> through publication." I wonder if an effort should be put forward to not just
> spell out Do's and Don'ts of Photoshop (what's been done), but also proactive
> Do's and Don'ts for conforming images to the printing press and other output
> devices.
>
> Arguments can be made to say that all presses are different and color tables of
> each must be had, but I take the point of view that a generic target can be
> developed and each printing press can then tweak.
>
> Comments?
>
> Jerry Sedgewick
> Diretor, Biomedical Image Processing Lab
> University of Minnesota
>
>
>
>
>
>
>
>
>
>
>
>
>
>
|
