About Banding Noise reduction (Canon)

[rodolgo]

Hello to everyone,

I'm shooting with a Canon 1100D (T3 in the US) and I start seeing noise banding on my shots when exposing over 60 seconds. I'm wondering what is the most appropriate step in the workflow to apply the banding reduction script, and to which frames : light, bias, darks, flats, before or after calibrating/stacking/noise reducting them?

Thanks in advance for your thoughts and precious experience,

Rodolphe

[georg.viehoever]

I usually apply after stacking. In theory, it would be better to do it on the RAW frames, but the reduction script currently does not support debayering (which would be necessary).
Georg

[burbs]

I've tried applying the script to the final integrated image and to the individual light frames (after dabayering, before registration).  There seems to be no noticeable visual difference between the two methods.

Its a lot easier and less time consuming to work on the integrated image rather than the light frames which require the image container to batch apply the script.  The only time it has been necessary for me to use the script on the individual light frames was when the frames were acquired on different nights and there was a reasonable amount of rotation between the sub frames.  Since the script only works on horizontal banding, if it was run on the integrated image there would be some residual banding going at an angle equal to the difference in rotation between the light frames.

[pfile]

burbs, i have run into the same thing. i even tried to rotate my integrated image by the required angle and re-running canonbandingreduction. it worked but not as well as if i had run it on all the subs.

i suppose it might be possible to hack georg's script to debayer. but it might be better to have the batchdebayer script be able to call georg's script. i assume one PJSR script can call another one...

to be honest after getting dithering going i have not seen the banding in the final result. while it makes sense to me that dithering could cure the banding, my tracking/guiding has always been bad and so it seems like the drift should have had the same effect as dithering. so its a bit of a mystery.

[georg.viehoever]

I had plans to translate the script into a proper PCL module, which would make it considerably faster, would make integration with other processes easier, would allow for the integration of other functionality. Unfortunately, I have not yet found the time to do it.

So if one of you guys feel like it: The algorithm is not really complicated. Go ahead, it is all open source, free to use, free to modify (as long as you make the result also free...).

Georg