March 12, 2016: 37-pane H-alpha mosaic after a lot of headaches

[michael.h.f.wilkinson]

This mosaic really gave me some headaches, as the flats seemed to cause no end of trouble. I must have stacked the data a dozen times with flats processed differently, before giving up. I got a rather blotchy disk with a clear gradient. Not good.

It might not look too bad, but any application of curves brought out the problems clearly. I then decided to do a little post-processing trickery. I first selected the disk in Gimp, with the ellipse tool, and mirrored it horizontally. I then loaded it, and the original in ImageJ, converting them to floating point. Adding the two images and dividing by 2 gave a disk without global gradient, but with a load of weird, symmetric detail. To get rid of the unwanted detail I smooth this result, and a copy of the original, by flat averaging with a disk of radius 200. I now had two blurry disks: one with the unwanted gradient and vertical banding, and one without. By dividing the latter with the former (after adding a constant to each pixel to avoid division by zero) gives me a synthetic "flat". Multiplying the image shown above with this flat, I get something much better, which I then could process further. The result isn't perfect, the part inverted ones show an artefact in the background, and I might tweak further, but at least it is a passable result.

Grey scale:

Part inverted:

Pseudo colour added:

Part inverted + pseudo colour:

Please do click on them and inspect the full size images

[alan4908]

Impressive images Michael. 

Alan

[Macavity]

Impressive. Have stolen quite a few ideas. Keep 'em coming!
As I am now discovering (with my Lunt LS50THa) *symmetry*
in such things is pretty difficult to acquire... I really value this.

My much vaunted centralised "sweet spot" has slipped a bit! 
Have decided to give a "tilter" and (a second) rotator a go...

Hoping I can optimise the hardware. Then use these software
ideas. Convinced there's a great scope struggling to get out?

[laudropb]

Very nice set of solar images. I think the part inverted brings out the detail well.

[David Smith]

Wow, quite a battle to get a decent image. Any idea what caused the problem?

Worth persevering, it would be a shame to just bin the data and your resulting image is none too shabby

[michael.h.f.wilkinson]

Cheers everyone!

8 hours ago, David Smith said:

Wow, quite a battle to get a decent image. Any idea what caused the problem?

Worth persevering, it would be a shame to just bin the data and your resulting image is none too shabby

I think the problem stemmed from the place on the disk where I took my flats. It wasn't centred well enough on the disk, so a residual gradient was present in the flat. I actually found that using a flat from a much earlier session worked a bit better.

[Pete Presland]

Nice end result despite the problems,  interesting thread discussion as well. 

[Davey-T]

Worth persevering Michael, I usually bin any images  that annoy me

Dave

[DrRobin]

That is a real shame about the flats, spoiling the full disc.  The filament/prom in the 11 o'clock position looks really spectacular, perhaps you could just process that frame and pull out a cracking shot? 

Robin

[Tyson M]

Nice images and processing breakdown. 

[Ewan]

Still a great end result Michael, feels even better when an image 'isn't all that' then to pull something out of the bag like that to get a finished result.

Great processing & capture.