What are these blue and white dots

Hi,

A thumbnail of the file can be seen here:

http://picasaweb.google.com/ronviers/Noise/photo#5040514739443530786

Where you can look at a small abstract of the EXIF data to the right
by using the 'more info' link. There are arrows pointing to instances
of the problem but you will need to use the 'Download Photo' link to
get a look at the dots. The link is to a 1.8Mb file so don't click on
it unless you really want to see.
This image has not been sharpened.

Thanks,
Ron

 

I was going to guess "noise", but they don't look like digital photo
noise to me in the big photo. There's also a lot of dust and stuff on
the actual microscope, it looks like, but I see a relatively few, fairly
big, blue and white spots in addition to the dust, and they are where
the arrows point, so I think I'm looking at the same problem you're
looking at.

Possibly hot pixels in your sensor? I haven't experienced that problem
so I'm not sure I know what they look like.

Any chance of editing mis-clicks or something? Looked like nothing
changed at the view level, but really a pixel or two got munged?

Or something I have no clue about.

I'll follow this thread with interest!

 

Hmm. I've never seen anything quite like that... But what's with the
weird exposure time - 717, or 17 seconds? Huh??

I'm also a little puzzled by the lack of any details to help
troubleshoot - do other images show this, or only this one? Have you
shot other stuff at long exposures? Do these spots always appear at
the same locations? Surely there are other clues..

By your description I was just expecting to see a few hot-pixels, but
they are just everywhere... And is this a 'straight' photo? or is
something else funny going on? The reason I ask is that you say there
is no sharpening, and yet the white spots often have severe ringing
artefacts as one might expect from oversharpening. The image itself
looks a little soft, yet many of the artefacts are very sharp indeed.

In the absence of a lot more info, I'm guessing some sort of
processing/firmware problem. Could there have been a lot of
electrical interference where this was taken?

 

At first I thought it might be small clumps of hot pixels. But
there are *many* smaller ones, and the color fringing along many of
the edge transitions between the microscope and the bright
background are also a similar blue. Did you notice that the blue
and white dots can be seen practically everywhere, including most of
the white background? My guess is that it's related to having such
a long exposure (717 seconds) and possibly the small f/22 aperture.
If you didn't already do it, could you take another shot using dark
frame subtraction to see if it makes any difference? I've read that
DSLRs have much better sensitivity than film, but sensors don't
handle extremely long exposures as well as film.

 

Might be hot pixels, which often show up as bright, colored blobs of blue,
green, etc.

 

Here is a dark version of the setup.

http://picasaweb.google.com/ronviers/Noise/photo#5040602781978131506

If you pin register the two you will notice that all the dots are
present and at the same locations in both images.

BTW, the images are unedited and the RAW settings were copied from the
ECE raw and pasted to the dark raw so it should be valid - of course
they are jpgs but the tiffs are too big to upload.

Thanks,
Ron

 

Thermal noise from a long exposure.
Using some HDR software, you can average or sum multiple, same, less long
exposures and effectively get rid of the noise.

 

My humble photo software doesn't do dark subtraction. I meant
that it might be useful to have the camera do the dark frame
subtraction. But as the dots are in the same positions, and the
exposure was the same, it seems that you've just gone beyond the
Rebel's ability to handle long exposures, at least at ambient
temperature used for the shots. Care to try again in a walk-in
freezer? <g> I'd guess that by getting into very long exposure
times your sensor's individual sensels' sensitivities aren't equally
linear, maybe due to nonuniform leakage. If anyone has better
suggestions for you and your camera it might be one of the several
rpd'ers more knowledgeable in long exposure astrophotography, R.
Clark's name coming to mind, among others.

And with my dialup connection, waaaay too big to download.
BTW, the dark frame is still downloading, but it has already
exceeded 1.5MB, close to the size of the original photo. Even with
the lowest jpeg compression ratio I'd have thought that a dark frame
would be highly compressible? It just finished, but I'm still
surprised that the 1,661kb dark image wasn't much smaller than the
1,909kb original. It guess that there's lots of very faint shadow
detail.

 

Thanks to everyone for their help. Capturing a dark image with that
setup was the definitive test. I was just pushing the sensor too far.
I reworked the lighting and got an acceptable shot staying within the
lens's comfort f range but loosing some interest in perspective. The
rules for this image require no out of camera processing so HDR and
noise reduction were not an option.
Those low light shots are tricky.

Brgds,
Ron

 

I'm still not sure you understood ASAAR's suggestion of using in camera
dark frame subtraction. Your reply was unclear if you actually used this.
Surely your Canon has this feature.

 

Not this model, that feature was added later. Even doing simple
subtraction in Photoshop made very little improvement (not that that
it is allowed) in the shadows and made the high key areas worse, so an
additional mask would be needed, because so little information
remained in the dots. I would need some third party NR software to try
to salvage the original setup.

Thanks,
Ron

 

That's sad. Even my old P&S CP-5700 had this feature, let alone my old D70.
I don't believe 3rd party noise reduction programs will deal with hot
pixels.