Thoughts on which imaging rigs to concentrate on

[Gina]

I'm trying to make a colour image from my three NB images using the Light Vortex Tutorials.  Firstly, preparing the images for combination.  Ran the Star Alignment - no problem.  Now I'm trying to do the Dynamic Crop.  Worst margins were on the SII image so used that to go by, dragged the instance to the desktop, then applied the process to the three images.  The result I got is shown below.  That is after applying Zoom to optimum fit for each image.

[Filroden]

You don't need to create copies of the process. Select your crop area on the target window but do not apply it. Then drag the half square to the other windows to apply that crop to them. Then apply the original crop using the green tick. 

[Gina]

Yes, that's what I did.  Tried again and it worked properly.

[Gina]

Went on to apply DBE which was successful then on to combine mono images to form a colour image.  Firstly, I tried using just Ha and OIII as the SII image is very poor but I don't think the result looks right.  I suspect the registration in Star Alignment.

Then equal H S O followed by SCNR as the image was very green.  The second image confirms the lack of registration I believe.

[Gina]

Back a few steps I guess to before Star Alignment.  That's all the processing I've done today!!!

[Gina]

Found my silly mistake!!  Picked up the wrong set of images after Dynamic Crop!  Now trying again.  I've checked that the images I'm using now are registered.  PI has the useful property of showing the top image partly transparent when moving it - it's possible to see the underneath image through it and see the images exactly line up as the frames are aligned   Clever stuff - the more I use PI the more I like it

DBE next...

Edited February 11, 2017 by Gina

[Gina]

Images look fine for registration but rather "washed out" - not much colour contrast.  Applied DBE and LinearFit, then combined colours with PixelMath. 

LH image is tri-colour using equal weights plus SCNR.  RH image is bi-colour using R=H, G=20%H+80%O and B=O.

Edited February 11, 2017 by Gina

[Gina]

Think that's enough for today   I'll look into other combinations in PixelMath tomorrow.

[Gina]

Couldn't resist one little tweak - applied curves manually instead of auto-stretch giving these images.

[Filroden]

I loved the tri-colour before you applied curves - there is a subtle shift between red and orange. But the yellow/gold after curves is amazing  Though I think you've raised the black point a touch too far and have lost some of the wispy nebula above the Heart that's much clearer in the first set of images. Are you applying masks when using curves?

[wimvb]

Great images, Gina, even if they're only screendumps. What causes the increased noise in the bi-colour (rh image)? Due to weak O signal?

[Gina]

4 hours ago, Filroden said:

I loved the tri-colour before you applied curves - there is a subtle shift between red and orange. But the yellow/gold after curves is amazing  Though I think you've raised the black point a touch too far and have lost some of the wispy nebula above the Heart that's much clearer in the first set of images. Are you applying masks when using curves?

Thanks   It was a very quick go with curves - I'll have another go.  No masks applied.  There's clearly a lot more that can be done with these images.

[Gina]

1 hour ago, wimvb said:

Great images, Gina, even if they're only screendumps. What causes the increased noise in the bi-colour (rh image)? Due to weak O signal?

Thank you   The O could have done with more data - yes it's a lot weaker than the H.  I'm still learning the best techniques with this combination of lenses and camera.  I think this season will probably be put down as learning and maybe next year I'll have a lot more experience and know more what to concentrate on.  Still hoping I can make some reasonable final images from the data I already have though.

Edited February 12, 2017 by Gina

[Gina]

Had another go with curves on the tri-colour but I think I've still cropped the black a bit.  Better but I'll probably another go.

Here's a full res version saved in PNG format.

[Gina]

Another go with curves on the bi-colour.

[Gina]

Bit too much black level on that one - here's another.  But trick to get the balance right but practice makes perfect they say   Hmmm... bit green this one - tried SCNR and it turned purple!!  (magenta)

Edited February 12, 2017 by Gina

[Gina]

Tweaked the green curve.

[Gina]

Been play around with different formulae in PixelMath but the only combination that seems to give any colour is HSO.  I guess the problem is that there's very little nebulosity showing in the SII.  Maybe I should stick to bi-colour until I can get more SII data.

[Gina]

Running the simulator I reckon if I got some clear night sky before too long I could capture two or three hours of SII for the Heart and Soul with the 135mm lens.  I could also do with more OIII.  Previous subs were captured with a gain of 440 but I could increase this to 500 or 600 and capture more data in a given time.  I think probably I'd best go for SII first as that's the worst set of data.

[Gina]

Just been reading about multiplying the Ha and OIII to produce a synthetic green instead of adding proportions of each.  This was using Photoshop but the same thing should be "do-able" in PI I think.

[Gina]

No, that doesn't work.

[Gina]

Used with other bits it did have a small effect.  I think I could go on playing with the formulae for ever and a day   For each colour the offset is changed by adding or subtracting a constant and the contrast is the multiplier.

Here's a screenshot of image and PixelMath.  Image was further changed with curves.

[wimvb]

4 hours ago, Gina said:

  Hmmm... bit green this one - tried SCNR and it turned purple!!  (magenta)

 

If scnr takes too much, I usually turn down the amount to .25 - .5. Scnr differs from mere curves tweaking by the way it treats highlights.

[wimvb]

2 hours ago, Gina said:

Just been reading about multiplying the Ha and OIII to produce a synthetic green instead of adding proportions of each.  This was using Photoshop but the same thing should be "do-able" in PI I think.

Can you post a link? (assuming you read it online)

TIA

[Gina]

Modified Bicolor Technique for combining Ha and OIII images