North America Nebula - Canon banding

[Ouroboros]

So far I've found the Canon banding utility in Pixinsight does a pretty good job at removing banding, but not with this image. See top right region.   I've tried varying the parameters in the utility and if anything less is more effective. But I'm always left with residual banding. Any suggestions?  Also, has anyone any idea what the vertical  line is at the bottom of the image I noticed on the last image I took to. It's on the original subs too, but not on the flats or bias.  

This compressed  image is a really quick process comprising: banding removal, dynamic crop, background neutralisation, colour calibration and basic stretch. Nothing fancy. It's 20 x 300s of dithered subs, processed with flats and bias. No darks. Camera: unmodified Canon 450D on a SW ED80 Pro. 

 

[wimvb]

CBR only works with strictly horizontal bands. Anything at an angle won't be corrected. If your subs show rotation (polar misalignment), CBR may not pick up the banding. One way to get around this is to use CBR on the individual subs. To use CBR in batch mode, load the subs into an image container, and apply CBR to that container. But do this before star alignment, since the alignment process will rotate subs.

Does the vertical line occur in all subs? if not, try removing it with large scale pixel rejection. Use a very clean sub as reference.

[Ouroboros]

20 hours ago, wimvb said:

CBR only works with strictly horizontal bands. Anything at an angle won't be corrected. If your subs show rotation (polar misalignment), CBR may not pick up the banding. One way to get around this is to use CBR on the individual subs. To use CBR in batch mode, load the subs into an image container, and apply CBR to that container. But do this before star alignment, since the alignment process will rotate subs.

Does the vertical line occur in all subs? if not, try removing it with large scale pixel rejection. Use a very clean sub as reference.

Thanks.  

Slightly confused by this. "But do this before star alignment, since the alignment process will rotate subs."   That reads like PS over-writes the original sub files. If so I'm surprised.  Or maybe PI creates new "rotated" subs. 

Actually I did the stacking in DSS so maybe I'm OK.  

 

[Ouroboros]

22 minutes ago, Ouroboros said:

Thanks.  

Slightly confused by this. "But do this before star alignment, since the alignment process will rotate subs."   That reads like PS over-writes the original sub files. If so I'm surprised.  Or maybe PI creates new "rotated" subs. 

Actually I did the stacking in DSS so maybe I'm OK.  

.... and ? How do you apply the Canon banding script to the image container?  Even highlighting the image container and starting CBR script gives error "No active window". 

[wimvb]

1 hour ago, Ouroboros said:

Thanks.  

Slightly confused by this. "But do this before star alignment, since the alignment process will rotate subs."   That reads like PS over-writes the original sub files. If so I'm surprised.  Or maybe PI creates new "rotated" subs. 

Actually I did the stacking in DSS so maybe I'm OK.  

 

Star alignment in PI creates new images, by default with "_r" added to their name.

40 minutes ago, Ouroboros said:

.... and ? How do you apply the Canon banding script to the image container?  Even highlighting the image container and starting CBR script gives error "No active window". 

Sorry for the confusion.

Star alignment not only shifts a sub to fit the reference image, it will also rotate the sub to get good alignment. This won't be by much, but may be enough to keep CBR from doing its work.

To apply CBR to a number of images:

1. load the images into an image container.

2. put CBR in a process container. The easiest way to do this is to apply CBR to any image, using the settings you want to use on the intended images. Then go to the process history of this "dummy" image and drag the small triangle over to the workspace. This will give you a process container (process icon) with the CBR process in it.

3. Drag the small triangle of the image container over to the process container. This will apply the process to all the images in the image container.

Hope this helps

[Ouroboros]

@wimvb Great. Got that.  The only prob' seems to be that my subs are CR2 files. PI doesn't seem to be capable of saving the processed files as CR2 image files. 

[Ouroboros]

Ah ... From what I read I don't think PI can save back as RAW CR2. It can open them though interestingly. 

I guess the only way to do this is to convert all my data to tiff files or whatever, apply the batch banding script to the lights and process (stack) in PI.  

[wimvb]

17 minutes ago, Ouroboros said:

The only prob' seems to be that my subs are CR2 files. PI doesn't seem to be capable of saving the processed files as CR2 image files.

That shouldn't matter. If you continue processing in PI, they will be xisf files. Otherwise you can save in fits format. Raw formats are not intended to be processed any further. They are intended as a fast and reliable way of saving image information from the sensor to the memory card, with minimal in-camera processing. Any software usually starts with converting raw files into a native format. Until fairly recently, the plugin of choice for many imaging packages was Dave Coffin's DCRAW.

1 minute ago, Ouroboros said:

I guess the only way to do this is to convert all my data to tiff files or whatever, apply the batch banding script to the lights and process (stack) in PI.  

Try this:

calibrate the images in PI batch preprocesser; select "calibrate only". Then deBayer; there is a batch deBayer process. Apply CBR to the lot. Star align and stack with image integration.

You can just put deBayer and CBR in one and the same process container. Then apply the image container that holds all the calibrated subs, to it. This will first deBayer and then correct the banding of each image.

[Ouroboros]

@wimvb ooh err!  Thanks for all that. I'll come back in about a week's time when I've done that.  So far I've only been toddling around in the foothills of PI - calibrating and stacking in DSS before using PI to process.   

[wimvb]

Sorry. PI is intimidating at first. But once you start to find your way around it, it gets easier. The image you posted originally was a bit on the bright side. If you slightly darken the background, you may be able to hide most of the offending bits.

[Ouroboros]

No need to apologise. It's a powerful program.  I had wondered whether I could hide the banding cosmetically. Unfortunately it occurs within some nebulosity rather than dark background. Cheers. Thanks for your help. ?

[Ouroboros]

.... A month later ... Just a follow up pic after learning some more about processing in Pixinsight. I did the calibration and stacking in PI. This did a better job than DSS IMO. I didn't actually apply the CBR to the individual subs in the end. The application of CBR after stacking, and the subsequent post processing, including darkening the background as suggested,  has got rid of much of it.