EAA post-processing

[AKB]

Let's face it, one of the nice things about EAA is that you don't have to go through hours and hours of data processing afterwards: you already have some really nice screenshots safely saved.

However, on those rare (ahem) cloudy nights, you might as well review the data you collected and see if there's more to be teased out.

I had a good evening of EAA on 2017-04-02, bagging (with Quattro 8 CF, HEQ-5 Pro, and Ultrastar, as usual, until the Hyperstar materialises):

• M101 (again) 22 x 1m
• NGC4631 The Whale (again) 20 x 1m
• Comet 41P (timelapse on a separate thread) 10 x 1m
• M3, 10 x 1m
• Two-thirds of the Leo Triplet (FOV not big enough!) 9 x 1m
• M104 Sombrero, 15 x 1m
• plus a tiny bit of the Elephant's Trunk nebula, 5 x 1m (too poor to show)

So I've done my best with reprocessing the FITS files, saved by Starlight Live, and done a bit of gentle processing (stretches, backgound levels, contrast enhancement, nothing fancy really) to see how the results might compare with the original screen captures.  I'm certainly no processing expert.  Some side-by-side comparisons below for your interest (hopefully.)

I think that in all cases the results are a tiny bit better, but really it goes to show just how amazing all the work that goes on behind the scenes in Starlight Live really is!

AK

M101

 

NGC4631

 

M3

 

LEO 2/3

 

M104

 

 

 

[Robrj]

Nice!  I've done similar things for showing them at home on the TV when friends visit.  What software do you use? 

[RobertI]

Interesting experiment. I think the difference on Leo 2/3 is most marked. I have tried doing some brief post processing on some of my captures and really couldn't make a huge amount of difference when changing curves, levels, etc, which i found pretty reassuring. The most marked improvement I got was when I did a quick sharpen on M57 (Ring Neb) and got significantly more detail out. Did you use anything like sharpening, de-speckling, etc? 

I think these experiments are very useful to help understand (a) how we could be using the EAA software better and (b) what enhancements to the EAA software may be beneficial. 

[AKB]

8 minutes ago, Robrj said:

I've done similar things for showing them at home on the TV when friends visit.  What software do you use?

In this case:

• Nebulosity for calibration and stacking.
• Matlab for home-grown 'pixel math'-type processing (background subrtraction, contrast improvement and noise reduction, here)

Both the above run on Mac and PC, and although I run the capture and mount on PC, I prefer the Mac for everyday processing.   However for other data, particularly solar system imaging, I use AS!2 and AstraImage which both require a PC.

[AKB]

26 minutes ago, RobertI said:

I think the difference on Leo 2/3 is most marked.

Yes, I'd agree with that, although M101 and NGC4631 are fairly good too.  Some of the improvement comes from dropping a couple of the poorer subs in the stacking.  I did get a satellite cutting the Whale in half in one...

 

26 minutes ago, RobertI said:

Did you use anything like sharpening, de-speckling, etc? 

'Sharpening' was a bit of home-made multi-scale contrast enhancement. 'De-speckling' was an implementation of @ollypenrice's curve flattening at the lower end (which I call ONR - Olly's Noise Reduction) so really just curves. 

No deconvolution or sophisticated noise-reduction was wielded in the making of these images!

[Paul81]

Superb comparison - this was one of my main aims with the FITS export to allow people to play with the data in a full-fat image processing application when the clouds have rolled in. I would always expect expert post processing could pull out more detail as there are limits of what you can do on the fly.

On the sharpening and de-noise front, I prototyped a wavelet based algorithm for SL at the tail end of last year and it works quite well and can make quite a remarkable difference. So this is something to look forward to in V4. Its a bigger change than anticipated as i realised i needed to change the whole concept of the user interface to accommodate it, as opposed to squeezing it onto another tab. The new UI has widgets for each tab that are in the current SL and these are dockable or floatable. This allows you to adjust wavelets and the normal display processing without having to keep switching. Well OK - it will do when I have finished it!! It also nicely solves the whole high DPI display issue reported too. I also have to optimise the wavelet code to make it more realtime so there is a lot of work to be done.

[RobertI]

4 hours ago, Paul81 said:

Superb comparison - this was one of my main aims with the FITS export to allow people to play with the data in a full-fat image processing application when the clouds have rolled in. I would always expect expert post processing could pull out more detail as there are limits of what you can do on the fly.

On the sharpening and de-noise front, I prototyped a wavelet based algorithm for SL at the tail end of last year and it works quite well and can make quite a remarkable difference. So this is something to look forward to in V4. Its a bigger change than anticipated as i realised i needed to change the whole concept of the user interface to accommodate it, as opposed to squeezing it onto another tab. The new UI has widgets for each tab that are in the current SL and these are dockable or floatable. This allows you to adjust wavelets and the normal display processing without having to keep switching. Well OK - it will do when I have finished it!! It also nicely solves the whole high DPI display issue reported too. I also have to optimise the wavelet code to make it more realtime so there is a lot of work to be done.

Wow, you have been busy Paul, lots for us to look forward to. Thank you.