Not improving

[Gavin1234]

Had a brief window of cloudless skies over the past two nights so I got some more data on my two current targets; horsehead and helix.

I’ve now got about 150 x 5 min subs of horsehead and about 380 x 5 min subs of helix. I’ve integrated the horsehead ones and it went down to about 100 frames during pre processing. Have I just stuffed up the post processing?

I’m still working on integrating the helix ones but I’m a little disappointed at the lack of improvement in my image with the amount of data I’ve been able to add. Truth is that if I was looking at them for the first time I’m not sure I could pick which one was first (with minimal data) and which one had all 100 subs. I know my colour balance is not the same but what do you guys think?

The first one with the very orange flame nebula is my first one with only about 20 subs. The middle one has a lot more subs and the final one with the blue stars and white flame nebula is my latest version with about 100 x 5 min subs used.

[glend]

Hard to know how to help without more detail. Your big stars are really burning up the images. What software are you using, what camera, scope, etc? Did you grade your subs? Using DSS and picking the best %?
Describe your workflow.

[The_bluester]

I think there is more to be had in the processing (Which I am a long, long way from expert at) but there is less noise and more detail in the last image than in the other two so the extra subs would look to be worthwhile.

[Gavin1234]

Thanks Glen, so I used a zwo071mc main camera cooled to -20. Sw Esprit100ED scope, CGX Mount. PHD2 for guiding with decent alignment. Sequence generator pro for exposures.

50 flats, 50 darks, 250 bias.

So using PixInsight I used blink to take out all the crap then I used subframe selector and took out the worst remainders based on FWHM then did the same again using SNRWeight. By the end of the time I stack the lights I’m left with 2/3rds of the originals subs i.e. 150 Dow to 100.


I’ve calibrated the darks and bias, then created masters before using them to calibrate the flats and then made a master flat. Then I calibrated the lights using the darks, bias and flat masters. Then I debayered them, then I aligned them, then I integrated them.

I usually do a small crop to start post processing, followed by an automatic background extraction, colour calibration, noise reduction using Atrous waveletTransform, Then I stretch with histogram transformation, then a curves transformation. That’s basically my workflow, but there are slight differences with each version for example in the first image I used tgvdenoise instead.

Thanks again for taking the time to help. I know none of these images are finished it’s just that I expected to see a much bigger difference at this stage in return for all the extra hours gathering data between the first image and the third.

[Gavin1234]

Thanks Paul, yeah I hope I can get more out of it with PixInsight (which I’m very new to). I’ll keep reading up on it but If anyone has any suggestions as to other PixInsight tools/processes I should run on these I’d love to hear them.

Thanks again

[Imme]

I disagree..... I think it is getting better!
I think the clarity is increasing and noise is reducing

[xelasnave]

There is a great improvement in my view.
Reprocess until you are happy.
alex

[sil]

you're headed in the right direction, but your workflow is still uniform to the frame, look into tutorials on enhancing nebulosity. I recall an awesome video by a german i think called Gerard? Which results in giving you a method to reduce those stars and bringing out their colours as well as being able to bring the nebulosity forward more (brighter/more contrast) without touching the stars. You workflow is never set or good, there is always something to improve.

I think youre third pic is a different colour profile too, i'm seeing artifacts typical of that. try making sure to convert to srgb before you send online.

[RickS]

Like others, I can see an improvement. You'll notice this more in the faint areas. The bright stuff gets to a high SNR with a small number of subs and then doesn't improve much more as you add data.



You don't need to filter out the low SNRWeight subs unless they are visibly cloud affected or otherwise damaged. By default, ImageIntegration will do noise weighting and give low weights to these subs but you'll still get some benefit from them.

Are you optimising the rejection settings when you integrate or just using the defaults? That might be a fairly simple way to get some more out of your data.



That's Gerald Wechselberger, aka "oldwexi". He has some good video tutorials on PixelMath and other topics: http://www.werbeagentur.org/oldwexi/PixInsight/PixInsight.html

Cheers,
Rick.

[rcheshire]

IMHO #2 is the right balance, showing lots of improvement and clean calibration. The others are a little overdone. The presence of the brighter stars is just right and the smoothness of the darker regions is a good indication. Nice job...

@multiweb - yes a little more saturation would make it.

[multiweb]

#2 with the color from #3. There is an improvement in your pics both in depth and colour balance.