Colour data processing in Pixinsight

[tooth_dr]

Can anyone share any decent tutorial for processing OSC in Pixinsight?  I've used Astro Pixel Processor to get the data calibrated, stacked, and the gradients removed.  I've opened the file in PI and done a Photometric Colour Calibration.  I reopen the image in APP and remove background colour noise using the HSL feature.  This leaves me with a linear OSC file that needs strecthed.  Where do I start with this is PI (if that's not a stupid question)

[scotty38]

Sounds like you know your way round PI so the easiest/quickest way imo is to apply an STF and then pass that to HT and then apply to the image as a permanent stretch.

Edited October 17, 2022 by scotty38

[tomato]

Until recently I was taking a calibrated, stacked and stretched OSC file into PI but after reading the notes on StarXtermintor I take a linear fits file into PI, separate the stars then perform numerous stretches on the starless image using Histogram and Curves Transformation. There is a script tool for stretches which reportedly gives better results but I haven’t got my head around that yet. I do a similar exercise on the stars only image, but a lot less radical. Recombine with Pixelmath using the RC recommended function, then apply NoiseXtermintor.

I have had mixed results with Photometric colour calibration, so this is not my default. I will use this, Colour calibration in PI or CSC in APP, or do it manually, depending on what looks ‘right’ to me.

[tooth_dr]

3 hours ago, scotty38 said:

Sounds like you know your way round PI so the easiest/quickest way imo is to apply an STF and then pass that to HT and then apply to the image as a permanent stretch.

I haven’t really much idea about PI. I just learnt how to do the PCC after another member on here showed me. I’m 95% PS for post processing but want to increase my usage of PI

[scotty38]

okey doke let me see if I can give you some bullet points but there are quite a few videos out there even PIs own channel if you want to have a look at some other stuff....

1. Open up the Screen Transfer Function process

2. Select the image you want to stretch and click the "radioactive" button in STF

3. All being well it looks ok but if it's too green for example uncheck the "link" icon top left corner and hit the radiaoctive button again

4. Open up the Histogram Transformation process

5. Pick the image to work on in the HT target drop down field just to make sure we're working on the right image

6. Drag the New Instance "Green Triangle" from the Screen Transfer Function on to the bottom bar of the Histogram Transformation process

7. You should see the histogram change accordingly.

8. Now drag the Green Triangle of the HT on to the image

9. If all is well the image should go white but don't worry that's because you've stretched it PLUS it has a screen transfer applied at the same time

10. Click the little computer icon on the right hand side of the STF process to cancel the STF and the image should go back to how it was.

11. You should notice the image now does NOT have a green bar at the side of its identifier

12. It is now a non linear image and can be saved/shared as such.

 

Hopefully I didn't miss anything and that is far easier to do than type 🙂 🙂 🙂 

 

I posted a link here to the PI channel.....

 

[teoria_del_big_bang]

13 minutes ago, scotty38 said:

okey doke let me see if I can give you some bullet points but there are quite a few videos out there even PIs own channel if you want to have a look at some other stuff....

1. Open up the Screen Transfer Function process

2. Select the image you want to stretch and click the "radioactive" button in STF

3. All being well it looks ok but if it's too green for example uncheck the "link" icon top left corner and hit the radiaoctive button again

4. Open up the Histogram Transformation process

5. Pick the image to work on in the HT target drop down field just to make sure we're working on the right image

6. Drag the New Instance "Green Triangle" from the Screen Transfer Function on to the bottom bar of the Histogram Transformation process

7. You should see the histogram change accordingly.

8. Now drag the Green Triangle of the HT on to the image

9. If all is well the image should go white but don't worry that's because you've stretched it PLUS it has a screen transfer applied at the same time

10. Click the little computer icon on the right hand side of the STF process to cancel the STF and the image should go back to how it was.

11. You should notice the image now does NOT have a green bar at the side of its identifier

12. It is now a non linear image and can be saved/shared as such.

 

Hopefully I didn't miss anything and that is far easier to do than type 🙂 🙂 🙂 

For the easiest method then that's spot on I think.
 

Steve

[wimvb]

Imho you shouldn't use the screen transfer function as a permanent stretch. The STF calculates the histogram stretch depending on image statistics. Its aim is to show what's in the image, without concern for aestethics.

Choose the stretch based on what you want the image to look like. With a colour image, I usually start with a colour preserving stretch, such as masked stretch or arcsinh stretch. I then apply a moderate histogram stretch, mainly to adjust the black level and increase contrast. After that, I only apply curves stretches. The exact choice of which stretch to use, and which settings for a certain stretch, should depend on the image.

[tooth_dr]

6 minutes ago, wimvb said:

Imho you shouldn't use the screen transfer function as a permanent stretch. The STF calculates the histogram stretch depending on image statistics. Its aim is to show what's in the image, without concern for aestethics.

Choose the stretch based on what you want the image to look like. With a colour image, I usually start with a colour preserving stretch, such as masked stretch or arcsinh stretch. I then apply a moderate histogram stretch, mainly to adjust the black level and increase contrast. After that, I only apply curves stretches. The exact choice of which stretch to use, and which settings for a certain stretch, should depend on the image.

I’m just starting out with arcsinh stretches in PS but wanted to try PI too. I can use DPP stretches in APP but they don’t preserve the colour data very well.

[teoria_del_big_bang]

18 minutes ago, wimvb said:

Imho you shouldn't use the screen transfer function as a permanent stretch. The STF calculates the histogram stretch depending on image statistics. Its aim is to show what's in the image, without concern for aestethics.

Choose the stretch based on what you want the image to look like. With a colour image, I usually start with a colour preserving stretch, such as masked stretch or arcsinh stretch. I then apply a moderate histogram stretch, mainly to adjust the black level and increase contrast. After that, I only apply curves stretches. The exact choice of which stretch to use, and which settings for a certain stretch, should depend on the image.

And for a proper workflow to get the utmost out of your data I would agree but when starting in PI the way @scotty38 suggests is a good way to start and like you say to see what is possible with the image.

Steve

[scotty38]

1 hour ago, wimvb said:

Imho you shouldn't use the screen transfer function as a permanent stretch. The STF calculates the histogram stretch depending on image statistics. Its aim is to show what's in the image, without concern for aestethics.

Choose the stretch based on what you want the image to look like. With a colour image, I usually start with a colour preserving stretch, such as masked stretch or arcsinh stretch. I then apply a moderate histogram stretch, mainly to adjust the black level and increase contrast. After that, I only apply curves stretches. The exact choice of which stretch to use, and which settings for a certain stretch, should depend on the image.

There are a million and one ways to stretch an image in PI so was just showing the simplest way to get one done

[tooth_dr]

8 minutes ago, scotty38 said:

There are a million and one ways to stretch an image in PI so was just showing the simplest way to get one done

I’ll try that out, but ultimately looking for a good way to do it, and just make an effort to get to grips with PI.

[tomato]

It’s easy to get intimidated by PI’s vast array of processes and scripts. I wouldn’t process an image now  without using PI at some point, but around 80% of the tools available currently remain untouched.