> What I was getting at is a perpetually confusing issue that most
>people don't understand, which is that an image taken with a digital
>camera has no size, no dimension. Or another way to say it is that
>the pixels have no dimensions. the various, loose uses of the word
>resolution are part of source of the confusion. the image just has a
>certain number of pixels or pieces of information. the image only has
>dimension when it is mapped to the pixels of the output device (let's
>not even talk about epson's 1440 - it's worse than MHz battles ;-).
>the distance thing was just a reasoning that if you didn't fill the
>frame with your gel, only a certain percentage of those couple of
>megapixels would contain useful information. a scan at 300 dpi would
>guarantee you 300 pieces of information for each inch of your gel.
>basically a red herring as you explained since my assumption that
>"more is better" is flawed.
>
>what about the issue of transmissive vs reflective imaging for
densitometry?
>
>I understand the 8-bit limit on screen displayed grayscale. is there
>any such limit (artificial or imposed) on the input devices? I guess
>this gets to your discussion on dynamic range and has to do with
>sensitivity on the input side. does it matter? another red herring?
>
>Please don't hesitate to post your response on the list. if you
>don't want to do it for some reason, could I anonymize it and put it
>out there?
Not posting to the full board was a mistake on my part, which I have now
corrected.
You are exactly right about the issues of resolution. I have a couple
Photoshop books that spend entire chapters trying to straighten out the
issues of what resolution are you talking about in which circumstances.
Anyway, your point is a good one. The camera image is (for example) 2k
pixels across, no matter how much of the frame is filled by the gel. If you
don't fill the frame you have wasted pixels you may need. On the other
hand, if you have a pro-level digital camera with a macro lens you could
focus in on a single band in the gel, and the image is still 2k pixels
across... probably more information than you can get from a 1000 dpi
scanner as far as resolution goes. Still, you only need enough resolution
to insure a good density measurement, and to fulfill your output
requirements.
I have never seen anyone try reflective imaging for densitometry of gels.
All photos of gels I have ever seen are taken on a lightbox from above, so
you are getting the light transmitted through the gel. I rather doubt that
reflected light would work.
As far as an 8-bit limit, that is a function of both the scanner/camera
analog to digital (A/D) converter and the software used. For example my
UMAX will output 10-bit greyscale, and Photoshop can use all 10 bits. Not
all software can, and not all hardware will output it. The newer scanners
have 14-bit A/Ds but will most often output only the "best" 12 bits. Still
that is a lot more information than 8-bits per channel.
The dynamic range I was talking about is not quite as simple as bit depth.
It has to do with sensitivity in the CCD, the electronics and the optical
path, as well as the number of bits the A/D converters divide the range
into. These factors can have a very large effect on the quality of scans,
particularly if you have high background levels or faint bands to pull out
of the noise. The more bits you can feed the software the more accurate
your data. If you have a large Dmax (density) you will also be able to read
a wider range if densities of the gel, where a scanner with a low Dmax
might give only noise at the extremes of lightness or darkness.
All of these issues are complicated to understand without having tried
them. If I hadn't spent much of the last two years trying to restore 150
year old family photographs I would never have gotten the hang of most of
them. I hope I clarified some of the issues, not obscured them...
Tim Slattery
Berlex Biosciences
This archive was generated by hypermail 2b29 : Thu Feb 01 2001 - 12:44:47 EST