- This topic has 8 replies, 3 voices, and was last updated 8 months, 3 weeks ago by
.
-
Posts
-
I have recently posted a couple of Ha images that I postprocessed with Pixinsight.
https://britastro.org/observations/observation.php?id=20240212_140055_6e0fd0b7b12b44cbThis is nominally a deep sky (DS) package but I used its gradient removal tool, and the ‘multiscale’ filter tool to remove residual (low frequency) Newtons rings by creating a synthetic ‘flat’.
I also tried its (AI-based) BlurExterminator tool. By turning off star sharpening, and winding down the non-stellar effect to around half maximum (and PSF of around 3.5) I obtained better results than with conventional Deconvolution (as you do with DS objects). I think the sun’s suface is not totally dissimilar to some deep sky nebulosity…Anyone tried this before?
Hi Kevin,
Yes, I would agree with you there. I’ve used BlurXterminator in a similar fashion and find that it also enhances fine detail in solar proms as well as surface detail.David
Here is a comparison image of large sunspot AR3576 right hand image processes with BlurXterminator.
Attachments:
It looks very nice Kevin. Have you tried ImPPG, a free resource available on github. Many solar imagers use this in preference to anything else, although I am not sure if they would have PixInsight.
There is a very nice deconvolution, tone/histogram tool and unsharp masking. They give excellent results I think, although I have not access to your tools in Pixinsight.
Kwong
I seem to recall giving imPPG a quick try…(?)
As far as I know, these tools all use the standard Lucy-Richardson (LR) deconvolution algorithm. The Xterminator suite of tools in Pixinsight only became available last year and uses a deep neural net (so-called ‘AI’ or machine learning) approach. My understanding is that the nets have been trained to recover Hubble images from deliberately corrupted versions therein, and are therefore sensitive to galactic and nebula-like features. These new tools were a game-changer in deep sky imaging; suddenly we could extract astounding levels of detail from our data.
While not tuned for solar work, they appear to work better (in my opnion) than standard LR. Unfortunately they are only availaibe in Pixinsight – which is not free. Maybe there is an opening for someone to repeat the AI approach for solar and planetary… 🙂Hi Kevin,
That would be interesting. Can I try your raw/stacked solar image and compare to you PI processed one ? That would be useful I think.
Not sure if you can give a dropbox or Google Drive link to that file ?Kwong
Here’s another Pixinsight add-on which looks useful, although I haven’t tried it yet:
Solar Toolbox Process for Pixinsighthttps://www.youtube.com/watch?v=3yzfH5x5Smg
David
- You must be logged in to reply to this topic.