Repair HSV & Masked Stretch

Scooot

Well-known member
When I attempt to use the repair HSV script to solve the problem of pink star cores following a masked stretch it never works.
See a couple of examples below.
Preview 2 before Repair HSV
Image 25 after Repair HSV
Masked stretch last

I used default settings on each and the unrepaired V

Can anyone help.
 

Attachments

  • DC_Preview02.png
    DC_Preview02.png
    37.2 KB · Views: 92
  • Image25.png
    Image25.png
    37.3 KB · Views: 87
  • Image25 masked stretch.png
    Image25 masked stretch.png
    378.3 KB · Views: 107
Hi,

You need to run the HSV script before the stretch. So, run the script "Repaired HSV Separation" with "V-No Repairs" checked.  This will generate 4 new images.  Now run the "ChannelCombination" process selecting the "Color Space" as "HSV" and add the "Channel / Source Images" for H, Sv and V".  Ensure that the "V" is the unRepaired_V" image.

Now you can stretch the image without the pink star cores appearing.

space is not black
John
 
Hi John, I did exactly that.

The first two images have just had stf appied so you can see the stars.
I then appied the script HSV script which generated the 4 images you refered to, and then used channel combination to generated the second image, (also visible due to STF).
I then used the mask stretch process to stretch it.
 
Images taken with a Canon 450DSLR
The work flow was as follows:

Bias integrated,
Darks calibrated & integrated
Flats calibrated & integrated
Lights Calibrated
Cosmetic Correction
Debayer
Sub Frame Selector
Star Align
local Normalisation
Integrate Lights (using Weights) then closed it without saving.
Drizzle Integration
Background Neutralisation
Colour Calibration
Repair HSV Script
Channel Combination
Masked Stretch
 
My workflow with DSLR is similar up to DrizzleIntegration.

Crop
Photometric Color Calibration
Dynamic Background Extraction
Canon Banding Removal Script
Multiscale Linear Transformation
HSV Separation Script
Channel Combination
Arcsinh Stretch

The image is now linear and further processing/noise reduction etc etc.

John





 
Hi John, yes mine is usually very similar with the exception of the Repair HSV because it never works so I use histogram transformation instead of masked stretch. I didn?t use DBE or any noise reduction this time because I?m solely processing to get to the bottom of why I can?t use HSV repair and masked stretch without pink star cores.
 
Hi again,

I assume that you reset the script and the process to their defaults first.  I have also set the "Repair Level" from 0.5 to 0.35 and also "Max Repair Radius" from 16 to 32.

Beyond that I am not sure ... perhaps one of the "Jedi Masters" have an idea or two ....

Look up
John
 
Yes I used the defaults and I also played around with the settings but it didn't seem to improve things, although I don't really know what i'm doing with the the HSV repair settings. How do I guage the required radius and degree of repair?

The read out for the two stars (which I assume are over saturated) is showing red 0.99.. whereas green & blue are much lower. Exposure was 90 seconds at F2.
 
Thanks for the original postings Mike, very interesting.

I don't think my issue is to do with working on a preview, I only used the previews for the purpose of posting this thread.

Interesting that the two stars prior to stretching don't seem to have a halo, the colour is pretty much red all over.
 
Here's a link to the whole image if anyone wants to try to use mask stretch without pink star cores.

https://www.dropbox.com/s/6pm5oq5kdimam5u/DC%20BN%20Col%20Cal.xisf?dl=0
 
Scooot said:
Here's a link to the whole image if anyone wants to try to use mask stretch without pink star cores.

https://www.dropbox.com/s/6pm5oq5kdimam5u/DC%20BN%20Col%20Cal.xisf?dl=0

Hi Richard - I downloaded your file and did a bit of processing with PI.

I believe the problem may be related to the alignment of your R/G/B channels to each other. I notice that you have what I call "Rainbow Stars" where the stars are red on one edge and blue or green on the opposite edge. I especially had this problem with my ES ED102. Because the colors aren't lined up properly the HSV can't correct both edges at once.

This isn't a very scientific answer, but, I believe it is definitely part (or possibly all) of your problem.

Hope this is helpful.

Steve
 
Thanks Steve, possibly an acquisition problem with my Samyang 135 F2 lens then? I thought my focus was quite good although the stars, (particularly the two I used in the crop) would have been better if I?d stopped it down a bit.

I?ll see if anyone else has anything to add but it seems I can?t do anything about it, just have to carry on using another stretch process.
 
Richard,

Are you using an OSC camera?
I'm not sure what I'm looking at in your first two pics. Are these the V component out of HSV repair? If so, they should be BW images. If you use OSC, then they probably need to be debayered, if you have not done so.
(As Steve noted, The third image looks as if the stars are red on one side and blue on the other. If there is a color registration problem, Repair may not work well, I'd suspect. Again, if using OSC, long exposures at low altitudes cause color dispersion that could cause the R-B variations across a star.)

Good luck, Alex
 
Hi Alex, sorry I?ve only just seen you?re reply.

Yes I?m using a canon 450dslr. The image was debayered during the calibration process.

The first pic is a crop of two stars, linear but auto stretched with Stf.
The second is the combined unrepaired v,s&h after running the script, again in linear and auto stretched so you can see it.

The third image is a masked stretch of the second imaged.
 
Back
Top