>   From: Barry J Burbach <[log in to unmask]>
>
> Continuing the discussion on z res, I get an elongation of Fluor beads mounted
> in oil. I'm assuming that it's because the ref index of latex beads is around
> 1.6 . I've calibrated my stepping motor by using 100 micron glass beads in oil
> with an oil lens in reflection mode. I seem to get no problems that I can
> perceive, although I seem to get a smaller focus step calibration factor than
> the manufacturer says ( Zeiss). I've tried to get flourescent methacrylate
> beads which would have a ref index closer to oil but I can't find any. With a
> cell in aq media and water lens, the images look pretty good until you go into
> the specimen a bit. I try to capture full range and threshold out values below
> the half max for the reconstruction. I figure that cutting the gain or
> screwing around with the black level isn't reality. I've been setting the step
> size equal to the FWHM, (my Optical sx) and interpolating out in the
> reconstructions. At z values much greater than x and y I get elongation
 
My philosophy is to take meaningful data in as large a quantity as possible
without running into problems such as bleaching, motion, memory space, and/or
data flow rates.  In general, you are doing the right thing by using
the full range of pixel values without a lot of saturation (to white or
black).  Once you have taken the data using this "linear" region of the
digitizer, you can do all of the non-linear processing that you want.
Having linearly acquired data makes things like deconvolution etc.
meaningful. One
problem is that you only have 8 bits/pixel or 256 gray values to work with on
the Zeiss (and most others as far as I know). I think a 12-16 bit digitizer
would be a great advantage, does anyone know if these exist on a confocal?
 
Using FWHM z-step sizes and interpolating data in between is certainly
reasonable, especially if you don't want a large quantity of data, but as I
mentioned above, if you can afford to take more data, take it. The
interpolation done by the scope will be better than anything that can be done
on the computer.
 
> problems simply from interpolation, and I may also have trouble from using
 some
> of the lower values. I don't see how you can have equal resolution in the XYZ
> by any method other than throwing away data that's real. In any case, I
> am well aware that Nyquist is turning in his grave,but I've been busy with
> other things. What's the deal?? should I be overlaping sections and scrunching
> them in the reconstruction? should I throw away my Z-res and not overlap?
 
I bet Nyquist is still resting easily because the system blur is acting as an
anti-aliasing filter for the step sizes you are describing.
Theoretically, you are always overlapping sections even when you sample at
more than FWHM (Nyquist twitches), so yes, take as much data as you can deal
with and overlap like crazy.
 
> I'd like to use that .7 or.8 resolution that we work so hard for. What's a
> biologist to do? Any of our resident heavies care to comment on any of these
> issues?
>                                 All the best,
>                                               BJ
>
 
 
Good Luck,
      Keith