Basque Portrait

cmr164

The character shows through... 100mm 1/20sec f4.5



The next has lots of blue channel noise (1998 imaging electronics) with the ISO 1600 and still 1/15sec 100mm f4.5 image stabilized

wxwax

That ISO 1600 really shows some noise, huh? Have you tried noise reducing software like Ninja Noise to see if it makes a difference?

cmr164

wxwax wrote:

That ISO 1600 really shows some noise, huh? Have you tried noise reducing software like Ninja Noise to see if it makes a difference?

I ought to write some s/w but I am too lazy. Or I should try some canned s/w for either UNIX or MacOSX.

I need to buy a new camera On the otherhand if I put the battery and the microdrive in my pocket and put the camera outside for 20 minutes in the -7F temp, the noise would clean right up. I have actually done that to get decent noise levels on 30 second new year's fireworks shots. (f16). In the 1985 timeframe I was using liquid nitrogen cooled sensors to make infrared images of exhaust plumes. Fun stuff.

wxwax

:eek1 No kidding? Cooling down the sensor reduces noise? Well, cut my legs off and call me shorty, I had no idea... and since you refer to -7, you must either be in Alaska or Maine.

EDIT - never mind, I see Waltham,MA on your work site. Plenty cold - I saw a TV weather reporter tonight in Boston open a bottle of water and pour it on a windshield - a good deal of it froze instantly. It's a bit nippy there tonight.

ian408

wxwax wrote:

:eek1 No kidding? Cooling down the sensor reduces noise? Well, cut my legs off and call me shorty, I had no idea... and since you refer to -7, you must either be in Alaska or Maine.

Astrophotographers use cooling for the longer exposures. But most use
specialized CCD cameras as well.

ian

cmr164

ian408 wrote:

Astrophotographers use cooling for the longer exposures. But most use
specialized CCD cameras as well.

ian

Well the liquid nitrogen cooled imagers that I worked with were for IR images of exhaust plumes and the exposures were very short. The principle is the same for all cases though. A cooled imager has way less spurious noise and is also better able to be affected by the heightened energy differential of incoming photons

jimf

cmr164 wrote:

I ought to write some s/w but I am too lazy. Or I should try some canned s/w for either UNIX or MacOSX.

I could write a filter for that kind of noise in a jiffy, I think; just take a 3x3 of the individual channel (I see noise in all three channels, FWIW) and look for a spiked value in the center; if you get one, average the surrounding pixels.

Crap, now you've got me thinking of writing image processing code again. I swear I just want to be a consumer this time around.

cmr164

jimf wrote:

I could write a filter for that kind of noise in a jiffy, I think; just take a 3x3 of the individual channel (I see noise in all three channels, FWIW) and look for a spiked value in the center; if you get one, average the surrounding pixels.

Crap, now you've got me thinking of writing image processing code again. I swear I just want to be a consumer this time around.

Will this do?

jimf

cmr164 wrote:

Will this do?

Quite :-). What did you use?

cmr164

jimf wrote:

Quite :-). What did you use?

'xv' (your nemesis ) has exactly the kind of filter you are describing. It does a Gausian blur only on the spiked pixels. It has an adjustable size matrix but I used 3x3 as you were suggesting.