I am trying to reduce the horizontal banding noise in my images. W10, MaximV6, AC4040 FSI connected through an ICRON Raven, CDK12.5 @2600mm, Baader edge painted CMOS filters, stack of 43 X 400s OIII, Gain 1650, internal stack size of 40, Bortle 4-5, 75% moon, all the necessary Flats and Darks, no GSENSE calibration. Please disregard the reflection pattern of the bright star which I believe come from my filters. I plan to flip them to test the result. They are present in the OIII & SII, but not the Ha for some reason. They also only appear in the stacked image. The banding is negligible on filters except OIII & SII, but on faint objects it is not. Note that LRGB images, and even Ha at approx. 4 X the signal of OIII are good. Following Doug's suggestion, I ran Band Reduction in Maxim V7 and present the resulting images: OIII stack, original OIII stack, Debanded-rotate 90R- Debanded- rotate 90L OIII individual files Debanded-rotate 90R- Debanded- rotate 90L, stack I think you will agree, that the Deband process worked well on the stacked version, but not well on the individual files / stacked image. Ultimately, I would like to correct the images at the source as much as possible to minimize the messing around and avoid losing detail. With my light polluted skies I plan to stick with Narrow Band imaging, so I want to optimize this rig for that. I plan to perform these tests when I can: 1) Reduce the High Gain from 1650 to something like 1400, and redo all the calibration masters 2) Take a set of images when the moon is not present 3) Take a set of images at 800 sec Any other suggestions would be appreciated, as well as any opinion concerning which of my tests are more likely to have a positive effect. Many thanks.
There are a few things you can do: Dither and stack. Normally you might dither by 2-3 pixels. Try going bigger, like 20 if you can. GSENSE Calibration. This involves taking flats at different ADU levels and interpolating. Very effective once you have it set up properly. We're still improving the UI on this tool. Filters / Band Reduction - do this on each individual frame. You can rotate 90 degrees and repeat (we're going to make vertical and horizontal built-in). You can use Batch Processing to apply this on all your calibrated images before you stack.
I have always dithered by +-20 pixels. GSENSE calibration has not worked for me as noted in that thread. Software fixes don’t cut it either. RGB images are much better than narrow band as noted above, but NB is what I want to do. I just finished testing it bin 2 and it is worse. I have a BSI and do not have this issue. I am wondering if this is not an issue with the electronics. This camera was repaired as it had a broken trace out of the box. Can it be tested under low light conditions (ie narrow band) by SBIG? I have run out of ideas and will be removing it from my observatory anyway.
I read the posting on the SBIG Blog concerning CMOS calibration which gave me some ideas, but I just can’t understand the huge difference in banding from my AC4040BSI (bortle 1) to my AC4040FSI (bortle 5) camera. I am dithering and calibrating the same way. Is it the sky or is it the camera? I repeat that the FSI camera is available for testing as it will be sitting on the shelf. Not usable.
Hey Mark - we're a bit swamped this week... sorry we're not as responsive as usual. Do you have the same brand of filters in each camera's filter wheel? Are they edge-blackened?
Do you have the RAW, unprocessed images that were used to make the file: masterLight_BIN-1_4096x4096_EXPOSURE-400.00s_FILTER-OIII_mono_MOSAIC-A1_autocrop.fit PixInsight trashes useful data and crunches the values to 0 to 1.0, making it really hard to analyze what is going on. I'm wondering if there is some problem light frame(s) that went into making that master.
Reminder: BSI - NM Bortle 1 dark skies Chroma - no banding but quadrants. FSI - QC Bortle 5 with moon Baader - banding and quadrants. If I recall there was a lot of variation between the frames. I am in New Mexico but I will try to upload them remotely and send you the link for both bin 1 and 2. I recall Doug telling me CMOS is sensitive to background more than CCD and I am not sure if that is playing a part. Also, the calibration frames were done before I read Doug’s article on your Blog …30min stabilisation, etc. … which should be compulsory reading BTW. I stumbled on it by accident but found it very useful. Another thing that Doug mentioned (NEAIC 2024) is that I should lower the Gain on the BSI to achieve less noise. Well I am experimenting now and lowered the Gain to 1500 and guess what? In the first processed image at least, the quadrants are GONE, without GSENSE calibration!!!! At the time he told me to leave the Gain on the FSI at 1650 but I am going to try lowering it when I get back along with the recommendations in the article.
Here is the link. https://can01.safelinks.protection....Q6TbGv8msHa2gAjBQT8pPjtTNPvhEw7mE=&reserved=0
Mark - I see a couple FITS keys i don't recognize: RDNOISE = 10.000000000000000 GAIN = 1.0000000000000000 Is that something you added?
No. Don’t they come from Maxim, or the GSENSE Calibration? I did try GSENSE Cal. I would have to check.
