How to use BlurXTerminator with AutoIntegrate?

Here is M51 with Crop, SPCC colour calibration, SXT, NXT and BXT, autoSTF TargetBackground = 0.10

Same image with a few small tweaks in Generic Processing:
(Increased saturation and maybe ABE and Local Hist Equalisation)

I’m not even sure if I prefer the edited version to the untouched one.

Hi Garth,

Nice to hear that you get good results! I think I need to add some configuration next.

The blank screen seem to be an issue with SpectrophotometricColorCalibration. I have reported it to PixInsight and they have reproduced the issue. Hopefully there will be a fix soon.

Jarmo

1 Like

Here’s M83 with the same settings as M51 and no additional edits.

Looks great, nice detail!

Jarmo

FYI - there is a BXT update 1.1.1 AND a new AI version 2

image

Here are the changes (from the documentation):

Don’t forget to open BXT in PixInsight and manually set the AI to V2

Good to know. I have not yet have time to try it. Does it work without problems in AutoIntegrate?

Jarmo

Worked fine for me with new version and AI V2

1 Like

Hi Jarmo,
I found some great information on BlurXTerminator recommended settings in this video for Version 1.1.1
The author communicates this via lengthy discussions with Russ Croman

The short story is the Workflow should strictly follow: SPCC > BXT > NXT all in linear phase.

BTW: I’m not suggesting that you need to do any of this…I’m just sharing because we have been searching for this kind of information.

Here is my summary:
(1:40) - resolution of stars is better in AI2 (previously 2 stars close together would appear as a single star…this is now fixed)
(2:00) - biggest improvement is No Color Change in Stars (AI1 would sometimes change star colors)
(2:35) - No significant change in sharpness of Non-Stellar Objects (sometimes better, sometimes worse)
(3:15) - New “Luminance Only” option - purpose is to ensure that star colors remain 100% the same but the trade-off is worse quality. The author recommends not using it. I think we would be wise not to use it given your enhancements around SPCC…any thoughts?
(4:50) - Clear recommendation on workflow sequence to apply BXT

  1. SPCC must be done prior to BXT. With AI1, SPCC could be used after BXT to correct the star colors problem in BXT V1, but this is no longer a problem with AI2. The biggest issue is that BXT changes the PSF structure of stars so SPCC may not recognize the stars correctly. Therefore it is recommended to run SPCC before BXT.
  2. (6:30) - NXT or Any DeNoise tool/process - Russ Croman makes a very clear statement that Noise Reduction should NEVER be done before BXT because you can never do any real deconvolution.…the information that SPCC needs is destroyed after denoising.
    **Note that the author is focussing on OSC primarily - but one of the comments addresses this and recommends that for multi-channel that denoise is postponed until after color combination, SPCC and BXT is done

(7:50) Auto PSF vs Manual PSF:
I’ve edited this section as the video was confusing but I have figured it out from the comments section and got Vastly Superior Results Here’s how it works:

  1. (8:20) Manual PSF is always better than Automated PSF BUT the tool will not correct star shapes if Manual PSF is used without the “Correct First” check box selected …I didn’t realize this.
  2. ((8:50) Russ’s recommendation is to calculate Manual PSF and enter this into BXT “PSF Diameter” AND select “Correct First” to get the optimal benefits from BXT.
    This will force BXT to: 1. Run Automatic PSF to get round stars; and 2. use the Manual PSF to do effective deconvolution.

    It sounds strange, but I have tested it on our M51 data and the results are much better…I will post these below
    (9:40) - Image Quality matters a lot…BXT gets far better results with good SNR Ratio…BXT does not invent new details out of noise.

M51 Test results for BXT AutoPSF vs Manual PSF with “Correct first” selected.
Same AutoIntegrate settings for both images:
Use SPCC for color calibration true
Background neutralization true
No noise reduction true
Crop to common area true
STF targetBackground 0.1

BXT applied to “autoRGB” after AutoIntegrate (so technically non-linear correct?).

Image 1: BXT with AutoPSF

Image 2: BXT with Manual PSF and “Correct First” selected

Both mages above with NXT applied manually after BXT:

Image 1: BXT with AutoPSF AND NXT

Image 2: BXT with Manual PSF and “Correct First” selected AND NXT

I will try this with M16 to test NB.
Thinking I would run the same AutoIntegrate settings (excluding SPCC - it doesn’t work for narrowband…Is that correct?)
I will manually apply BXT and NXT after AutoIntegrate (as I did with M51).
Do you think this is the right approach?

M16
Background neutralization true
No noise reduction true
Crop to common area true

BXT applied to AutoRGB after AutoIntegrate.

Image 1: AutoPSF

Image 2: Manual PSF and “Correct First”

Tried it on the linear image “Integration_RGB” for fun, applied BXT ManualPSF; NXT; ABE; HistogramTransformation; Hue shift for more orange; Hue shift for SHO; Remove Green Cast:

Zoom:

Hi Garth,

Lots of good into and image to check! It will take some but really good info.

I have been running with SPCC > BXT > NXT so good to know that is the recommended way.

And you are correct, SPCC is not run on narrowband images in AutoIntegrate. I checked some documentation and my understanding was that SPCC would work for HOO images but not for example with SHO images.

I am right now struggling with plate solving in AutoIntegrate. Cannot understand why it does not work there while running ImageSolver separately works :slight_smile:.

Jarmo

1 Like

Hi Jarmo,
I can see that you have been really busy.
I must say that your SPCC feature is fantastic.

I don’t know much about image solver…are you using it to automatically identify the object in SPCC?

Sounds very odd that it will work alone but not inside the script.

Please let me know if you would like me to run any testing, only if it helps you, otherwise I will give you some space. More than happy to help with anything in any way that I can.

Garth

Image Solver Script failures:

Like I said, it’s a bit over my head, but I did read this post where Image Solver was failing because the coordinates were not getting passed over.
The problem had to do with the image NOT being the active window.

I know it’s a long-shot, but I was wondering if, when running a script, the focus is NOT on the active window vs when running manually, it may be the active window?

Also, happy to run a test on my side to ensure it’s not something environment specific?
I can install the same AutoIntegrate version, use the same data (which I can download) and the same settings (which you could provide.
Sorry to keep bothering you - no need to reply if you don’t want me to do this.

Hi Garth,

Extra testing is always good! But I do not have a specific tests in mind. So whatever you feel like testing or just processing is good. Like those posts that you did are useful to see how different options change end results.

My problems with plate solving were related to data that does not have coordinates embedded or even a correct focal length. Now I have focal length as a configuration parameter and then I needed to calculate resolution from there. After that it worked ok but it took some time to realize :slight_smile:

I have an automated test setup (originally done by Jean-Marc Lugrin) which has some problems to test with SPCC because the dialog does bad. So again something that I do not fully understand :slight_smile:

Jarmo

1 Like

Hi Jarmo,
Glad to see that you got it sorted.
In hindsight, my last post didn’t make much sense because your issue wasn’t environmental as it worked when you ran it manually.

Regarding testing, I will continue to update AutoIntegrate when I see new releases and run against test data to try to identify any Mac specific issues.
I will also use the latest version for my own processing, which may help identify issues outside of our test data.
I will also work through various scenarios and paste images for comparison. On this one - please feel free to mention any specific scenarios that you may want me to test if the need arises.