Can this image be salvaged?

[sreilly]

I have uploaded a zip file that contains the rgb image of vdB1 taken last night. The file is about 24 MBs and can be downloaded from http://www.astral-imaging.com/vdB1-9-rgb_c.zip . Caph is just off the chip, in this case to the upper left corner. The negative affect is causing all kinds of problems as you can expect at 2.27 mag. I tried using DBE but suspect it may well be beyond it's ability or I'm simply not using the tool correctly. Normally I have no issue using DBE but then again I don't usually have such bright stars just off the image frame. Any suggestion would be great. The images is acquire using my OGS 12.5" RC and SBIG STL-11000M camera with Baader 2" unmounted filters. The image is 9 hours of combined data with each frame being 20 minute each at -30C and binned 2x2. Combined using Sigma Clipping, averaged, clipping high/low pixels/range. Star aligned and clean edge cropped.

If there's a way to make a decent image of this data I'd love to hear how you do it.

-Steve

[Carlos Milovic]

Hi Steve
I would just try to use DBE to substract that huge halo (not division). With linear, non-saturated data, this should work quite well. You may end with problems with saturated stars. In that case, the only thing that I can recommend is using a mask build from the luminance. I fear that the colors of the stars may still be wrong, but this should be better than opaque stars.
Oh, btw, try multiple iterations of DBE to really fine-tune the results. Use the substraction operation, using the normalization option.

[sreilly]

Carlos,

Thanks for the suggestion but let me make sure I understand. Where should I take my samples? Just in the halo corner or everywhere else? I was actually using both subtraction and division but will correct that. Making a mask for the saturated stars should be easy if I follow this right as the standard mask settings usually pick up the brightest stars for me. Are you saying to use this mask with DBE? I usually have my DBE Target Image Correction settings with Normalize, Discard background model, and Replace target image all checked.

I really appreciate the assistance. I usually have great success with PI and have made it my only processing program. I still use PS to make my different sized jpg images for my website as it makes quick work of resizing the image and file size depending on quality settings. I suspect I can do this as well in PI but haven't spent any real time looking at this. Still trying to improve my image processing skills in PI.

My latest posted images processed in PI are found here if you care to look. http://www.astral-imaging.com/latest_images.htm

Thanks Again

[Carlos Milovic]

Steve,

Yes, you are using it right. Samples should be quite dense in the halo, and a few outside, to keep the background flat and avoid oscillations on the synthetic model. You may need to repeat DBE one or more times, to get the result you desire. To simplify this, just create a process icon right before you finish your first DBE. After applying, just double-click the icon, and make any needed changes for the second/further iterations.
You should use subtraction, since this is an additive effect (light from the bright star is added to the background sky and the stars). The normalization option should ensure that the background remains at the same level, and that you effectively only removes the halo.

[sreilly]

Carlos,

I tried to take sample points in the halo area but they can't be used (red). Only very much outside the halo can I get good sample points (green). When I tried just the halo area I got a message that I needed more sample points. Only two outside were green and PI said I needed at least 3. There must have been at least 50 inside the halo area not being used.

Thanks for the advice.

[Carlos Milovic]

Increase the Tolerance parameter to the maximum allowed, and decrease the shadows relaxation (to 1). Now you should have more samples. Also, try disabling the weights (enable "Unweighted").

[Harry page]

Hi

Had a quick go and you can remove the halo, but had to up the tollerance up to 4 and do 5 dbe runs  , also I lowered the smoothing factor to o.2

To get a good result lots of fine tuning may be required , but can be done

harry

[sreilly]

OK, now I'm just starting to feel more dumber. If I increase the tolerance to say 3 and set the shadows to 1 I get really bad color distortions. That's with and without the weights enabled.

[Harry page]

Hi

I would only use the normaliztion ( or not at all ) after the first run of DBE and I tried to remove the halo only during the first few runs and then I did the fine settings for the last run

Harry

[martin_magnan]

Hi

I have tried your image with dbe

Here is the result with the xpsm file used for dbe. Just be carefull and flip your image vertically (vertical mirror) before using the icon.
Only one iteration is necessary

http://dl.dropbox.com/u/15597168/Dbe.zip

Let me know what you think

Martin

[sreilly]

Looks great but I haven't a clue what the xpsm file is used for. Is this a process icon and if so how would I use it? Thanks for the assist.

-Steve

[sreilly]

After dragging the file to the PI desktop it became very apparent. Thanks I'll give this a go.

[sreilly]

I just played with the data using the process icon from above and learned a good deal about the DBE tool. This is a much stronger tool than I had previously thought and well worth an in depth tutorial dealing with various types of situations. I'm thinking that this tool may be able to reduce some of my previous halos where one color comes through stronger than the others. In particular the blue channel seems to be more problematic on very bright stars then the other filters.

Thank you very much Martin for the assistance. I think I may have learned something already today!

[martin_magnan]

Hi Steve

you are welcome

I strongly encourage you to play with the settings, in particular the tolerance and the smoothing factor. Tolerance can probably be lower than what is in the process icon. I usually use the lowest working Tolerance and the highest working Smoothing factor.

Martin

[Harry page]

Hi

Martin gave some good advice , but using a smoothing factor of 0 is a very rare occurance and can cause problems  but as said play away 

Harry