crab nebula, how to improve the image?

[iwols]

hi all just trying pixinsight with a different work flow ,not loads of data( about 20 mins of each lrgb)  ,how would you improve this please,and how do you spread out amount of l against rgb thanks

Edited December 16, 2021 by iwols

[rl]

I would defer to the imaging experts on this one but I don't think that is a bad image considering the aperture and integration time involved. There is quite a lot of structure visible in the nebula. The pulsar is one of two very close faint stars in the middle; they are not quite split. The Crab is quite a small deep-sky object to image well with a small scope. 

Everyone is going to say "get more data" which is statistically true for random noise. But there is a law of diminishing returns that sets in; you have to keep increasing the integration time out of proportion for the same step of improvement. 20 mins is a bit short. A couple of hours on one object is usually feasible in one session for most people leading normal working lives! Technically, the signal/noise ratio is a Poisson distribution if you're interested in the maths. the signal/noise improves as the square root of integration time.

Was it taken on the transparent night with poor seeing last weekend (10th dec?). Good seeing is worth a lot. Mine taken on that occasion with an 8" scope look a bit better but not much, with 2 hours data. Your stars look a bit blobby but that might be down to over-enlargement. Did you use a Bahtinov mask?. If you're guiding with PHD2 you can see the guiding errors and how they compare to normal.

I can't see too many hot pixels so presumably you did a dark frame subtraction. But there is some noise on the left side that might want some sort of gradient correction. 

I think the colour balance is not far from the truth. The colours in the Crab are quite subdued unless you force the saturation in the processing. I use Siril which does colour balance by adjusting the colour ratios to match the spectra of stars in the image. It plate-solves the image and goes to an online catalogue of spectra before tweaking the colours for the best spectral match. Siril will do the gradient correction mentioned above as well. And it's free (but you are at liberty to make a donation!)

There are some real experts on this site ( I am not amongst them!) Hopefully one of them will chip in and give you chapter and verse concerning resolution and pixel size....but I don't think you are far wrong in that respect. 

We could use some more info on how the image was obtained. Guiding, focussing, Seeing, processing et cetera. 

Edited December 16, 2021 by rl
factual error

[iwols]

 

3 hours ago, rl said:

I would defer to the imaging experts on this one but I don't think that is a bad image considering the aperture and integration time involved. There is quite a lot of structure visible in the nebula. The pulsar is one of two very close faint stars in the middle; they are not quite split. The Crab is quite a small deep-sky object to image well with a small scope. 

Everyone is going to say "get more data" which is statistically true for random noise. But there is a law of diminishing returns that sets in; you have to keep increasing the integration time out of proportion for the same step of improvement. 20 mins is a bit short. A couple of hours on one object is usually feasible in one session for most people leading normal working lives! Technically, the signal/noise ratio is a Poisson distribution if you're interested in the maths. the signal/noise improves as the square root of integration time.

Was it taken on the transparent night with poor seeing last weekend (10th dec?). Good seeing is worth a lot. Mine taken on that occasion with an 8" scope look a bit better but not much, with 2 hours data. Your stars look a bit blobby but that might be down to over-enlargement. Did you use a Bahtinov mask?. If you're guiding with PHD2 you can see the guiding errors and how they compare to normal.

I can't see too many hot pixels so presumably you did a dark frame subtraction. But there is some noise on the left side that might want some sort of gradient correction. 

I think the colour balance is not far from the truth. The colours in the Crab are quite subdued unless you force the saturation in the processing. I use Siril which does colour balance by adjusting the colour ratios to match the spectra of stars in the image. It plate-solves the image and goes to an online catalogue of spectra before tweaking the colours for the best spectral match. Siril will do the gradient correction mentioned above as well. And it's free (but you are at liberty to make a donation!)

There are some real experts on this site ( I am not amongst them!) Hopefully one of them will chip in and give you chapter and verse concerning resolution and pixel size....but I don't think you are far wrong in that respect. 

We could use some more info on how the image was obtained. Guiding, focussing, Seeing, processing et cetera. 

hi thanks,c8 edge 0.7x reducer on a heq5 ,414 ex camera with filter wheel guided,last night with 3/4 moon,processed in pi cheers

[AKB]

I am not an expert by any means, but I was interested to look at your data.

A commented above, it's a bit hard to say too much without some processing details.  If you wanted some detailed input from experts, then I think that posting a raw stack FITS file would be the way to go.

I see the following things:

• background – indeed,  there is a noticeable gradient which should be removed early on to help with the rest of the processing
• green cast – easily removed with SCNR
• noise – some from the limited data, no doubt, but some also introduced, or exacerbated, by something in your processing... sharpening or filtering?
• colour saturation – your choice, but it's quite a muted palette
• stars – seem quite large.  This may be focus, or stretching.

Just my thoughts.

Tony

 

[iwols]

hi thanks link to original fits thanks 

https://www.dropbox.com/sh/m2c35zd70ifcac9/AACnbNMtQT5IWvfhk7QKAOaaa?dl=0

[AKB]

Thanks for that.  Here's my quick attempt.  I clipped off the edges since the registered frames are fairly scattered about.

Not so different from yours, perhaps, but I've tried to address some of the things on my list.

Not so much data there, so more would help, and focus could, perhaps, be a bit sharper?

 

Edited December 16, 2021 by AKB

[iwols]

thanks akb must admit your attempt is quite a bit better than mine and definately better stars ,regarding focus it was bang on,,its a problem i seem to have with my c8 edge ,hoping the reducer was going to make it better but stars still seem soft,thanks for the image any way appreciated 👍

Edited December 17, 2021 by iwols

[iwols]

what was your workflow Tony?

[AKB]

10 hours ago, iwols said:

what was your workflow Tony?

Yes, fair question!  I didn't put any details earlier in case you weren't really interested in following up.

I processed this with an L-RGB approach, keeping the  L stack separate initially, and combining the RGB ones.

L processing:

• large crop around edges to remove stacking artefacts
• DBE to remove background
• TGV noise reduction – you have only very little data.
• Histogram/Curvers – stretch, but not too much
• StarXTerminator – to separate nebula and stars
• Starless image – stretch and Local Histogram Equalisation to enhance contrast
• PixelMath – to recombine with stars

RGB processing:

• Noise reduction
• stretch
• SCNR to remove green cast

L-RGB combination:

• register RGB with L
• use LRGB to combine

That's about it really.

Since you have so few frames, there is nothing clever you can do in the stacking to remove noise even better (Sigma clip, and the like.)  More frames would definitely help here.  One things very noticeable is that your separate frames are of very different quality, and some would have been rejected if more were available.  Here's a couple of L frames cropped:

 

 

 

Was this, perhaps, high level cloud?  How good is the guiding?

HTH

Tony

[AKB]

One final go – I tried a little harder by combining the L channel with a synthetic L from the RGB combination.

So really you've got something good going there – just need more data and careful processing.

Tony

 

[iwols]

well Tony that last image is a world apart from mine,thanks,regarding hi level cloud must admit these lights were taken with a 3/4 moon,one last thing, i understand your workflow except register RGB with L.Thanks for your input and time highly appreciated👍

[AKB]

Oh, well I had cropped the L early on, but not the RGB, so this was simply to apply the same crop to RGB before LRGB combination. 

[iwols]

Thanks Tony will have to look into the StarXTerminator as ive never used this before,watched literally  hours if not days of pixinsight tutorials but never got totally to grips with pi over the last few years,ive had some good images but mostly luck than actually having the perfect workflow,and i also find my C8 images more challenging to process than my ts optics images.