NGC 6946 AI vs Manual Processing (Vitali)

NGC 6946 AI vs Manual Processing

I've tried to use BlurXTerminator (BXT) and NoiseXTerminator (NXT) AI-based tools in my PixInsight processing workflow. Revision B shows the result of manual deconvolution and de-noising. Then I've re-processed the data again from scratch and replaced deconvolution with BXT and all de-noising operations with NXT to make the original image. I've tried to keep all other processing steps identical (stretch, color calibration and saturation, etc).

BXT settings were: Sharpen Start 0.25, Sharpen Nonstellar 0.35. The latter value is much lower than the default 0.90. Results with higher values looked too artificial to me

NXT settings: Denoise 0.50, Detail: 0.15.

BXT did a great job on the stars. I usually apply a tiny amount of MorphologicalTransformation to slightly shrink the stars. But BXT was able to make them much smaller.

I've noticed one issue with BXT: sometimes some small structures in a galaxy, for example bright star clouds, are becoming much smaller than they should be. Probably BXT thinks that they are stars(?). Deconvolution does not produce such effect. Please see the revision D and its description, which shows two small clouds in UGC5829. It looks to me that the right cloud has become smaller compared to what I would expect.

Thanks for looking!
Vitali

B

Title: Manual Processing

Description: My best attempt :)

Uploaded: ...

C

Title: Animation. AI version has smaller stars.

Uploaded: ...

D

Title: Deconv vs BXT on UGC5829

Description: Arrow points at two small clouds. The left one is much smaller than the right. Also SDSS image shows this. Deconvolution result keeps the left cloud small. It looks to me that BlurXTerminator makes the size of these two clouds almost similar, the left cloud is just a bit smaller than the right one.

Uploaded: ...

E

Title: Crop of AI version

Description: Tweaked stretch a bit and cropped.

Uploaded: ...

F

Title: Side by side comparison of the source image and both processing workflows

Uploaded: ...

G

Description: Using latest BlurXTerminator and NoiseXTerminator tools.

Uploaded: ...

PixelSkiesAstro

	Distinct awarded users	Total awarded images
Image of the day
Top picks
Top pick nominations

	Image of the day	Top pick	Top pick nominations	Total submitted
Deep sky
Solar system
Extremely wide field
Star trails
Northern lights
Noctilucent clouds
Landscape

	Image of the day	Top pick	Top pick nominations	Total submitted
Backyard
Traveller
Own remote observatory
Amateur hosting facility
Public amaeteur data
Professional, scientific grade data
Mix of multiple sources
Other
Unknown

Submitted The date and time when you submitted this image for IOTD/TP consideration.	...
Views by Submitters (available since September 19th, 2023) Every image is assigned to 50% of available Submitters. In the event that at least 80% of them don't view the image before its time in the IOTD/TP process expires, it's assigned to the other 50% of Submitters and the process begins anew.
Promotions by Submitters When 3 distinct Submitters promote the image, it moves on to the next stage of the process: evaluation for Top Pick status. This requirement, in addition to anonymization of images and distribution to only a subset of them, prevents biases and ensures that the best images are selected.
Promotions by Reviewers When 3 distinct Reviewers promote the image, it moves on to the next stage of the process: evaluation for IOTD status.
Early dismissal Staff members have a lot of images to inspect on a daily basis, and they can dismiss images if they believe they don't meet the requirements for IOTD/TP selection. If an image is dismissed 5 times, it's removed from the process. This streamlines the process and ensures that any bias present in promotions could be overruled by other staff members.

Status	Advanced success
Started	...
Astrometry.net job	9053732
PixInsight job	KIYYOCAJHIKF6H1HDT0F6YE47YH8APVR
PixInsight queue size	n/a
PixInsight stage	TASK_LOG

RA (center)	20h34m50s.664
RA (top/left)	20h37m45s.934
RA (top/right)	20h31m57s.606
RA (bottom/right)	20h31m58s.211
RA (bottom/left)	20h37m41s.780

Dec (center)	+60°09′04″.67
Dec (top/left)	+60°22′47″.90
Dec (top/right)	+60°23′14″.60
Dec (bottom/right)	+59°55′06″.73
Dec (bottom/left)	+59°54′42″.57

Image Index:

Contribution Index (beta):