I am currently going through the process of trying to get a good process of my panorama. Had another go at it this afternoon, this is a 10x9.
Issues that I am aware of are:
Banding - Trying to figure out how to remove this, having serious difficulty in figuring this out. The banding is imprinted into the background and at different levels in each frame depending on how close the signal is to getting over the read noise (banding is created on sensor readout).
There is also some banding moving through the middle of the image, same thing. Low signal area; lots of black.
Satellites - This is my very last thing I will be removing. Going to do it on the small handful of frames individually as a final step.
Blue cast - (along the lower two panels and in the middle a little) This is a consequence of the banding, the lower three panels (I haven't included the first panel at all) which have the lowest signal of all of them. Will be dropping the lowest and maybe both, not sure yet.
As a final attempt I may end up trying to rectify this in PS but that won't be on the individual images like everything else.
What I am unsure of:
Mostly colour at this stage. I was at a friends place last night for dinner and asked his opinion (on my non colour calibrated laptop) on colour when I was doing some pre-processing testing. He has no idea on astro colour btw, I was mostly just asking whether the upper region was golden speckle without any random colour casts
I still want to go and attempt to do some contrast enhancement on the black regions (dust) to make them pop more but one thing at a time. Just needing some clarification to see if I am on the right path now before moving further on in processing.
Can't really nit-pick, but it's a fine rich image. Maybe a little dark IMHO? For me, the satellites are not a big issue, though I understand what you mean about the blue cast. I'm not sure how to tackle that. But a view well worth exploring!
Can't really nit-pick, but it's a fine rich image. Maybe a little dark IMHO? For me, the satellites are not a big issue, though I understand what you mean about the blue cast. I'm not sure how to tackle that. But a view well worth exploring!
There is a banding correction tool in Carboni Photoshop actions. Not sure how well it works but it could be worth a try.
Is it possible its a result of poor stitching? What stitching program did you use? Pt Gui is unlikely to do that.
Greg.
Ill have to give it a look, I have tried the DBE tool in PI with limited success. It is visible in the individual frames so into is definitely not a stitching issue. The banding isn't occurring where the stitches are.
Using AutoPano Giga at the moment, I've tried a trial version of PT GUI but it hadn't been nearly as good at doing the job.
Ill have to give it a look, I have tried the DBE tool in PI with limited success. It is visible in the individual frames so into is definitely not a stitching issue. The banding isn't occurring where the stitches are.
Using AutoPano Giga at the moment, I've tried a trial version of PT GUI but it hadn't been nearly as good at doing the job.
I wouldn't worry too much about it. Skyglow, different conditions at the time of imaging and the wide area covered all conributes to this.
Not sure if it will give better results for you but I've been very impressed with the stitching (Photomerge) in the later versions of Photoshop. It's under: File>Automate>Photomerge
Lightroom now also has similar functionality but I've never used that for astro stuff.
I actually had a crack at your jpg the other night in PS, i'll post it later i am not sure if it is as good as your new version though (different screen).
I wouldn't worry too much about it. Skyglow, different conditions at the time of imaging and the wide area covered all conributes to this.
Quote:
Originally Posted by pluto
Not sure if it will give better results for you but I've been very impressed with the stitching (Photomerge) in the later versions of Photoshop. It's under: File>Automate>Photomerge
Lightroom now also has similar functionality but I've never used that for astro stuff.
Want to try to get rid of the banding if I can, will look considerably better when printed on my wall This picture here is a single frame of a low signal area. With the push of a button AutoPano Giga is able to automatically align typically about 95% of the frames saving a LOT of time. I am pretty staggered at how well it does at frame alignment all on its own.
Quote:
Originally Posted by rustigsmed
hi col,
I actually had a crack at your jpg the other night in PS, i'll post it later i am not sure if it is as good as your new version though (different screen).
cheers
russ
Would be interesting in seeing it Russell. My new one was redone from scratch so it has a better black and colour calibration.
There is a technique I use sometimes with varying degrees of success:
- Pick an area of your image that resembles what you want your background to look like and sample with the eye dropper (at least 5x5)
- Create a new document the same size as your image and populate it using the paint bucket tool
- Paste this document as a layer on your original image with blending mode "lighten"
- Adjust opacity as required
GoodLook 64's Displayer has a "Corduroy" button which does a pretty good job at removing patterned read noise banding. It has to be done on the individual frames before assembling the mosaic, because in general the bands won't line up from frame to frame.
The bands need to be horizontal. If they're vertical, rotate the image 90 deg, then de-corduroy, then rotate back.
Be interested to know if it does the trick on your image, but it should. Doing 90 panels one at a time could require some fortitude, but each panel is very quick.
There is a technique I use sometimes with varying degrees of success:
- Pick an area of your image that resembles what you want your background to look like and sample with the eye dropper (at least 5x5)
- Create a new document the same size as your image and populate it using the paint bucket tool
- Paste this document as a layer on your original image with blending mode "lighten"
- Adjust opacity as required
This may help with the banding.
Cheers
Steve
What I had originally in mind was using PS and selecting the banded regions (the bluer ones) and just selectively playing with the colour. Not sure if my computer will like doing stuff to a 17gb .psd file More food for thought though.
Quote:
Originally Posted by Placidus
Not sure that there's much to fix!
GoodLook 64's Displayer has a "Corduroy" button which does a pretty good job at removing patterned read noise banding. It has to be done on the individual frames before assembling the mosaic, because in general the bands won't line up from frame to frame.
The bands need to be horizontal. If they're vertical, rotate the image 90 deg, then de-corduroy, then rotate back.
Be interested to know if it does the trick on your image, but it should. Doing 90 panels one at a time could require some fortitude, but each panel is very quick.
Best,
Mike
Should this be done on the linear data (as the very first process) or at the end after manipulation?
make of it as you will. it was done pretty rough and quick - thus why i hadn't uploaded.- i also was sloppy around the cats paw.
i think you are better off making it a bit more contrasty in the darker areas it helps with the banding.
pretty much selected the banded areas and used selective colour sliders.
also increased contrast and i think slightly moved the black point to the right in levels.
Should this be done on the linear data (as the very first process) or at the end after manipulation?
Shouldn't matter a great deal. What it actually does:
- On each scan line, finds all data fainter than a specified centile brightness. (there's a slider, but the default works well). This helps it ignore stars and bright nebulosity, and concentrate on the background.
- Calculates the brightness of the background on that scan line.
- Bandpass filters the resultant background brightness to get a smoothed estimate of the corduroy pattern.
- Subtracts that from the image.
So it doesn't matter so very much when it does that. But I tend to do it earlier rather than later.
make of it as you will. it was done pretty rough and quick - thus why i hadn't uploaded.- i also was sloppy around the cats paw.
i think you are better off making it a bit more contrasty in the darker areas it helps with the banding.
pretty much selected the banded areas and used selective colour sliders.
also increased contrast and i think slightly moved the black point to the right in levels.
your original is attached on the right.
cheers
Increasing the contrast certainly helps. I've added one that I did a week ago on the original (the 200kb version) in Instagram.
Quote:
Originally Posted by Placidus
Shouldn't matter a great deal. What it actually does:
- On each scan line, finds all data fainter than a specified centile brightness. (there's a slider, but the default works well). This helps it ignore stars and bright nebulosity, and concentrate on the background.
- Calculates the brightness of the background on that scan line.
- Bandpass filters the resultant background brightness to get a smoothed estimate of the corduroy pattern.
- Subtracts that from the image.
So it doesn't matter so very much when it does that. But I tend to do it earlier rather than later.
I tried it, it looked like it was doing something but didn't seem to have any effect on the large banding. Maybe the banding is too large to get an effective background bias?
Looks very promising, Colin. Have you tried the CanonBandingReduction script in PI?
Cheers,
Rick.
I hadn't thought of that! Just put in a 15 minute test (did it on 90 edited frames) and it shows some promise BUT I'll have to put in the hard yard and spend a few hours doing it to the linear data.
Banding - Trying to figure out how to remove this, having serious difficulty in figuring this out. The banding is imprinted into the background and at different levels in each frame depending on how close the signal is to getting over the read noise (banding is created on sensor readout).
There is also some banding moving through the middle of the image, same thing. Low signal area; lots of black.
Hi Col,
Not sure if this will meet your need, however wanted to post just in case. A program I use specifically for removing banding in images is called "Fitswork". It's free and you can view/download it here - http://www.fitswork.de/software/softw_en.php. I have no affiliation with the software, I'm just a happy end-user.
Fitswork is very good. It also does other astro processing things, however all I do is run the program, open my image (usually the final stacked image, prior to processing) and run the following menu command: Processing, Background Flatten, Lines to Equal Values. I then typically use an Absolute Median Level of 50.0 (although you can alter it to achieve the desired effect).
I've attached a sample of how it helped me remove banding from a 15min exposure on my DSLR (note - ignore the star trailing, I was purely testing the noise/banding issues that might crop up when using such a long exposure). I also ran your noise panel through Fitswork and have attached the end result. It looks like it could be making a positive difference.
On a final note, if you have banding issues in your individual frames, could you be multiplying the issue as you mosaic the whole thing together?
Not sure if this will meet your need, however wanted to post just in case. A program I use specifically for removing banding in images is called "Fitswork". It's free and you can view/download it here - http://www.fitswork.de/software/softw_en.php. I have no affiliation with the software, I'm just a happy end-user.
Fitswork is very good. It also does other astro processing things, however all I do is run the program, open my image (usually the final stacked image, prior to processing) and run the following menu command: Processing, Background Flatten, Lines to Equal Values. I then typically use an Absolute Median Level of 50.0 (although you can alter it to achieve the desired effect).
I've attached a sample of how it helped me remove banding from a 15min exposure on my DSLR (note - ignore the star trailing, I was purely testing the noise/banding issues that might crop up when using such a long exposure). I also ran your noise panel through Fitswork and have attached the end result. It looks like it could be making a positive difference.
On a final note, if you have banding issues in your individual frames, could you be multiplying the issue as you mosaic the whole thing together?
Cheers, Evan
Thanks for that Evan, I'll have to give it a look over. It appears like it does the same routine as the CanonBanding routine in PixInsight. A quick test suggested to me that it should never be done on edited data I have started reprocessing every frame making that my first step. Couple of hours in it appears to be making a difference although now that I am nearing the end (20 more to go, do that tomorrow) it looks like it has thrown my whole colour balance off and damaged some other aspects so I think I'll have to start over again anyway :/
More food for thought though.
