Urocentrum

discomorphella · Post by **discomorphella** » Sat Sep 15, 2012 1:04 pm

I have not had a chance to get to my pond in quite a while, but I finally took a walk over and among other creatures, I found this little guy zooming rapidly around. Of course, I had to try and capture him, but Urocentrum moves very rapidly, so these pictures are the results of frantically focussing and shooting. By using flash synch at 1/320s on my D300 I managed to get a few shots at high magnification before he ran off. Optics: BX-60, DIC with 60X/1.2 NA objective, U-DICT upper Nomarski prism, U-DCA set at 2.0, Leitz Variozoom eyepiece (set at 5X) and 0.32X reducing lens relay to Nikon D300. Flash illumination using a SB-800 flash which is right now sitting on a petri dish behind the lamp house, shining directly through the 100W halogen lamp with the lamp house cover removed. Right now I am using the BX-60's internal ND filters and also changing the SB-800's intensity to change exposure, which is not optimal. I also need to machine a better mounting for the flashgun.

David

dennisua · Post by **dennisua** » Sat Sep 15, 2012 1:17 pm

Nice images, though there are lots of camera noise. Where that comes from? Not fully exposed or something else?

discomorphella · Post by **discomorphella** » Sat Sep 15, 2012 1:19 pm

Not sure where the noise originates from. It is almost certainly from some setting on the D300 or post-processing.

David

Charles Krebs · Post by **Charles Krebs** » Sat Sep 15, 2012 2:31 pm

David,

Your optics are obviously working beautifully on what is a very difficult subject to photograph alive! Great details.

I too am puzzled by the noise... actually not so much by the overall level, but by the "unevenness". It is much heavier across the top of the frame (where there is also a faint magenta band). I'm puzzled why you used the higher 1/320 sec "FP" sync, while setting power levels on the flash manually. Is there any reason to do so? (Naturally the fasted sync speed possible is desirable, but I was under the impression this was primarily to use "pulsed" flash for high shutter speed sync). I've never tried this myself, and wonder if it may be a reason for the odd upper edge noise/streak.

U-DCA set at 2.0, Leitz Variozoom eyepiece (set at 5X) and 0.32X reducing lens relay

So does that translate into a 3.2X mag of the objective... i.e. 180X on sensor.

rjlittlefield · Post by **rjlittlefield** » Sat Sep 15, 2012 4:02 pm

Any chance that the top of the image was originally much darker and has been mostly fixed up by a flat-field correction?

--Rik

dennisua · Post by **dennisua** » Sat Sep 15, 2012 4:06 pm

Sorry for OT but how do you make flat fields with microscope?

discomorphella · Post by **discomorphella** » Sat Sep 15, 2012 5:18 pm

Rik is correct, these images were corrected by a background divide using imageJ. And after a bit of debugging it looks like there is something wrong with my imageJ corrections; I don't see the noise band when I view the images using other software. I may have to switch to Matlab, or finally get Photoshop or similar for flat-field and other corrections.

David

Edit: for dennisua, the way I have implemented a flat field correction is to divide the image pixel-by-pixel with a blank image. That normalizes the background intensity and also removes dust and other artifacts to some extent. This is easy to do using imageJ's calculator plugin, but I am afraid it's also introducing noise as well

Edit: Charlie, yes the total mag on post objective is ~3.2. this is clearly into empty mag a bit, but I wanted to oversample a bit before I downsampled to 1024 by 680

rjlittlefield · Post by **rjlittlefield** » Sat Sep 15, 2012 5:52 pm

denisua, the phrase "flat-field correction" refers to correcting for nonuniform brightness by locally adjusting the luminosity curves. See http://en.wikipedia.org/wiki/Flat-field_correction for a more complete discussion.

Dividing by the gray levels from an image of a blank slide is one simple approach that usually works pretty well.

Even if the correction is not buggy, it can increase apparent noise if the original images are locally dark, using only a small portion of the gray scale in some areas. In this case the noise is really present in the inputs, but it gets amplified by the division.

--Rik

discomorphella · Post by **discomorphella** » Sat Sep 15, 2012 6:10 pm

Hi Rik-

I think you are correct that the flat field division is part of my noise problem. Perhaps not all of it though, since I am not getting such severe amplification using a different bit of code. For some reason the images look very noisy in imageJ but not other applications, even before being divided. I will continue to debug this, and post an example corrected image using different software.

David

dennisua · Post by **dennisua** » Sun Sep 16, 2012 3:49 am

David,
Do you mind posting an original image (before all the processing) and flat field somewhere on the web?
BTW. You also need to subtract bias image (shortest exposure possible with no light entering camera) from flat field to make it work correctly.