fringe killer filter add blue back to central star

[admin]

Sorry about opening this old post, but I tried the trick I did in this post (by happy-kat » Sun Apr 22, 2018 8:24 pm) using 1.6.686 and I could not get the same effect correction on the stars, I was using other data. and in the Decon module I see there's been a change in naming. Previously
Parameter [Mask Behavior] set to [De-ring Mask Gaps, Show Result]
Is this the same as Enhanced Deringing and if the slider is to the left is that same as the previous [De-ring Mask Gaps, Show Result] please?

Setting Enhanced Deringing to Off indeed reinstates Decon's "old" behaviour. The main issue with using deconvolution to address chromatic aberration is that it can only work correctly for stars that don't overexpose (e.g. don't have "fat white cores"). Chromatic aberration is most visible precisely around overexposing stars, so that makes using this technique tricky in these situations.

It is also worth mentioning that chromatic aberration in datasets that have been color balanced and - worse - camera matrix corrected will have the chromatic aberration bleeding into the other color channels, making it much harder/impossible to use channel-based deconvolution to address this issue.

[happy-kat]

http://forum.startools.org/viewtopic.ph ... t=20#p5540
I am trying to use this trick above in 1.6.386 or 1.6.387 and when I have reached a point of Compose and bringing my new blue channel into the red and green and get to the Colour module I am getting a very blue image and I unable to colour balance it, the stars are effectively ringed in blue or green depending on how I try to balance it. I have tried with or without interpolation in the compose module on the blue image.

So any help please on is it still possible in theory to use the decon module as I had done in the link above to tweak just the blue channel image or is it as above not a route to take. I can use the filter fringe killer I just wanted to exhaust the other option to either use it or put it to bed and move on.

[admin]

http://forum.startools.org/viewtopic.ph ... t=20#p5540
I am trying to use this trick above in 1.6.386 or 1.6.387 and when I have reached a point of Compose and bringing my new blue channel into the red and green and get to the Colour module I am getting a very blue image and I unable to colour balance it, the stars are effectively ringed in blue or green depending on how I try to balance it. I have tried with or without interpolation in the compose module on the blue image.

So any help please on is it still possible in theory to use the decon module as I had done in the link above to tweak just the blue channel image or is it as above not a route to take. I can use the filter fringe killer I just wanted to exhaust the other option to either use it or put it to bed and move on.

It may be possible in theory, though it depends on the quality of the dataset and whether the stars you are trying to fix are over-exposing or not.
Perhaps you can share a (crop of) a linear dataset and I can see what the best approach may be?

[happy-kat]

binned 50% linear not white balanced

Autosave002 bin50hcp.tiff (298.98 KiB) Viewed 9003 times

Linear no white balance 1100d 50% binned heavy crop and some example offending stars and a bit of nebula.
Old 200m lens with CA.
I have a process using filter fringe killer but just trying to exhaust the processing approach methods.
I have seen some other tips of your's on another thread so off to read those.

[admin]

Autosave002 bin50hcp.tiff

Linear no white balance 1100d 50% binned heavy crop and some example offending stars and a bit of nebula.
Old 200m lens with CA.
I have a process using filter fringe killer but just trying to exhaust the processing approach methods.
I have seen some other tips of your's on another thread so off to read those.

Thank you for uploading. I was kind of hoping for a colour image (so the re-combine can be evaluated) though the chromatic aberration halos in this example seem very disfigured to begin with (which will make it hard to use Decon for this purpose).
That said, the new Decon module (of which 1.6.387b gives you an early taste) will include a special mode that is specifically designed to correct for some focus issues (chromatic aberration could be thought of as a subset of that). Again, this would still require "round"-ish halos to begin with. Stay tuned for this functionality in 1.6.388!

[happy-kat]

Sounds interesting, 1.6+ has been superb at bringing out data I did not know I had captured.

[admin]

Hi all,

If you find one channel is particularly unreliable/usuable (such as the blue channel in the case of certain types of chromatic aberration), you can also just decide to leave it out altogether for the purpose of your detail/luminance (but not for your colouring!).

To do that, launch the Compose module, load the same dataset for red, green and blue, set "Luminance, Color" to "L + Synthetic L From RGB, RGB". As you (hopefully) know by now this will let you process luminance and colour separately, yet in parallel. This mode will also let us synthesise a synthetic luminance dataset, weighted according to the Exposure sliders, while using the colours as-they-are for the color dataset.
E.g. we can now create a luminance/detail dataset that is decoupled from the colouring. And that's exactly what we'll do;

We'll set Blue Total Exposure to 0 (e.g. "Not set"). Now the detail/luminance dataset will not use any detail from the blue channel, however the colouring information to colour the remaining detail will still use the blue channel as-is.
Because you did not white-balance the dataset, set the Green Total Exposure to double the exposure amount of the Red Total Exposure (to reflect you have twice the amount of green samples per 2x2 Bayer matrix patch vs only one red sample).

Now process as normal. You should notice the halos caused by the blue channel are much diminished, while colouring is still entirely correct.

Hope this helps!

[happy-kat]

Thank you Ivo.
The final process has tighter stars which look much better, star colour is good but my Orion is now much blander in coluring, The CA is much less evident after filter module use though I had a few rings that crept in along the way.
I will put that data set down for a bit and lay with other data I have.

[happy-kat]

I have been playing today and this is my process using the Compost module.
The positives where I tweaked the process to remove the CA colour bloating by using a mask in Colour module that excluded the very large CA stars and their bloat.
The down side was I could not tease the colour out and M42 is bland.
Whilst I loaded blue as 0% I forgot to load green as 200%.

NewComposite autosave002 v8 cp.jpg (118.33 KiB) Viewed 8927 times

So I took the mask concept for the Colour module and processed as normal for OSC and no white balance and is linear.
This has much better colour on the stars and M42, and I controlled the CA as before.
The downside is the Compose process does has slightly more contained stars on the extreme bloated stars.
note downsizing and saving as jpg has added star halo artifacts that are not present in the tiff.

Autosave002 v9 cp.jpg (109.13 KiB) Viewed 8927 times

I wonder could we add as an idea please, an option in the Compose module on selecting Keep of adding to the two options a third of OSC no white balance and is linear as per is we had opened the file straight. OR maybe the option would be better in the colour Module or perhaps I need to play more with the Colour module if I have used Comopse with OSC data from a DSLR.

[happy-kat]

Did a new easier way last night so may revisit this other data set above to try it there.
Bin | crop | wipe | autodev | HDR etc. | mask set to exclude stars | colour background and nebula | mask set on stars but excluding all fat chromatic aberration inflected stars | colour the good shaped stars | mask set to fat chromatic aberrated stars | tracking off | shrink stars
This process leaves the chromatic aberration rich stars as luminance mono and looks a lot better.