[RCC] Markarian's Chain Requests for constructive critique · Jure Menart · ... · 6 · 301 · 1

Dear forum,

I'd like to ask for constructive criticism on my latest image: Markarian's Chain & friends

This was intended to be mosaic of 4 panels, but this year I was able to get only these two panels (mosaic should extend to cover also region of M89 & M90 but this might be work for next year).

I know I could use slightly more denoising, but I decided against it to keep it a little bit 'rougher' when zoomed in. I am hanging 50:50 for and against this decision. I am also aware of incorrect shape of NGC 4473. This processing was actually 'quick' processing done (I didn't intend it to be final), but private things didn't allow me to reprocess it a month ago. I will probably start from scratch processing now and this time I plan to use drizzled data, for this reason I think RCC is just perfect as I could take any comments for new process of the image. I plan to try to stretch & process M87 to show relativistic jet on full image. Will see if this will look OK or not.

As mentioned on my image, the main two goals were to show the M87 relativistic jet, which I showed with separate processing, it is possible. Another goal was to show also hydrogen alpha clouds NGC 4438 & M 86 - this goal is not achieved as Ha data (even though it's 25 hours) was acquired under extremely bad conditions. Last goal of course was to try first time mosaic and to make generally 'nice' image.

Processing steps in PixInsight:
- WBPP was used to prepare master files (with darks, flats and dark flats)
- Dynamic crop of all master data
- RGB channel combination
- Creating super luminance from L + R + G + B (with ImageIntegration) - I actually noticed yesterdat that super-Lum has lower SNR compared to Lum, so maybe I will avoid this or try to prepare it differently
- RGB & Super-Lum background extraction - I used multiple iterations of DBE (some iterations were quite aggressive to remove light leak and because data was acquired under Moon shine) - at the end I was quite happy with background
- Adding Ha to RGB pane4 (I had Ha data only for this pane) - I used HaIsolation and HaRGBComb PixelMath scripts, I am always using. Note that Ha signal was very weak, but I was able to get some additonal Ha data for some smaller galaxies, so I decided to add it anyhow. I also know this is not normal place to add it (usually I add it after SPCC), but I decided to add it before doing mosaic
- Create mosaic for HaRGB and Super-Lum (ImageSolver, MosaicByCoordinates, PhotometricMosaic - I didn't use TrimMosaicTile script as I didn't see the need for it)
- RGB - run SPCC with usual sensor & filter settings, set background neutralization on non-galaxy part of the image
- BlurX on HaRGB (sharpen nonstellar 0.40) and Super-Lum (sharpen nonstellar 0.80), other values default
- NoiseX (denoise 0.50 or 0.60 - can't remember, detail 0.15) on Super-Lum
- StarX on Super-Lum and HaRGB
- Starless HaRGB: stretch with HistogramTransformation & STF helping
- Starless HaRGB: small ColorSaturation & Curves to gently pull out some colors (using mask for brighter parts so the color in background was not affected)
- Starless Super-Lum: stretch with GHS in couple of iterations, trying to focus not to saturate the galaxy cores
- Starless Super-Lum: HDRMultiscaleTransform - can't really remember settings, but I was aiming to have brightest parts in range of 0.8, was quite happy with the details in galaxies
- Starless LRGBCombination - needed few iterations to be sure that colors are not killed by too high luminance
- Starless HaLRGB - ColorCaturation & Curves with few iterations to pull out colors as much as possible (using mask for brighter parts so the color in background was not affected)
- optional (can't remember if it was applied or not this time): Starless HaLRGB - UnsharpMask
- Stars RGB - HistogramTransformation
- Stars RGB - Curves to try to put some more colors from stars
- Combine starless and stars: PixelMath script I am always using, using the combine() and op_screen()
- Saved as JPG with 95% quality factor so the size was appropriate for astrobin upload

I like the overall composition and the idea of trying to get the whole chain in.

In order to show the M87 jet you have had to compromise on showing the extent of it's halo – which then makes M87 look relatively smaller than it should be whereas in reality it is the dominant object.

One way around that would be to do a high dynamic range composition of the M87 segment –effectively applying a lesser stretch to the core of M87 than to the periphery.  That way you could still have the brighter jet shing through and yet still show M87 as bigger than any of the other galaxies.

Just a quick comment about the Ha.  Is there a red shift problem  reducing the signal ?  If your filter is too narrow (3 nm)  then at 50 M ly  the Ha may be shifted up far enough that  the peak is only at the edge of the bandpass.  This maybe a case where a 7 nm or wider filter would be better?



Tim

Tim Hawkes:
I like the overall composition and the idea of trying to get the whole chain in.

In order to show the M87 jet you have had to compromise on showing the extent of it's halo -- which then makes M87 look relatively smaller than it should be whereas in reality it is the dominant object.

One way around that would be to do a high dynamic range composition of the M87 segment --effectively applying a lesser stretch to the core of M87 than to the periphery.  That way you could still have the brighter jet shing through and yet still show M87 as bigger than any of the other galaxies.

Just a quick comment about the Ha.  Is there a red shift problem  reducing the signal ?  If your filter is too narrow (3 nm)  then at 50 M ly  the Ha may be shifted up far enough that  the peak is only at the edge of the bandpass.  This maybe a case where a 7 nm or wider filter would be better?



Tim

Hi Tim

Thanks for the comment.

I agree the M87 + relativistic jet on wide-field image might be too much. I want to try it and see if I can make it nice (I see the data is there). So this is for me experimental (I also didn't do yet so extreme dynamic range stretching) and would like to try it and just revert to 'normal' stretching if I can't make it work.

Regarding the Ha: It's 6.5nm filter, I don't believe the red shift is the cause. The cause is IMHO that I took big majority of this data with full Moon and target was very close to the Moon itself I will try to gather this again next week under more appropriate conditions (~10-15 degrees from Full Moon ).

andrea tasselli:

Jure Menart:
Hi Andrea

Thanks for your feedback.

1. Are you referring to noise with 'rather grainy'? Do you think this is removed if I run denoise at final stage as shown below?
2. I tend to agree - I think this is result of (over)saturation and pulling out the blueish channel. At the same time I actually like more yellowish galaxies on this image. I probably pushed it a little bit over the edge. As I also see too satured/unnatural yellow/blueish chuncks  around bigger galaxies (see the denoise example below - maybe this is the grainy part you are referring to?)
3. Stars colours - I agree, I think 180s is a little bit too long for my current setup (I compared them to 60s acquisitions and I like those better). I will be more careful with stars preparation (maybe do additional integration of just best acquisitions and next year maybe even re-take only stars)
4. stretching - do you have any specific example of stretching issue? I tried to be careful with Super-Lum stretching but maybe I could try to pull out more details?

Hi Jure,

1. I assume that the image scale would have required far more exposure than you actually got in the end so basically you image is data-starved at the mid-tones/shadows level. That's what grainy (thus noisy at a larger scale) is meant to mean. A far less aggressive stretching putting your highlights at a lower intensity level would ameliorate the situation and allow some more degree of de-noising whilst avoiding the flattened look that over-smoothed images have.

2. Yes, probably toning down the saturation point and trying not to over-push the blue tones will give a more "natural" look.

3. I would have to disagree about this point. I'm using the same telescope and I have no problem in retaining star colours with 300s exposures. Different imager but you should still have enough dynamic range for your focal ratio to retain colour information in all but the very brightest stars with 3 min exposures.

4. See my point (1.) above. You can only stretch as much as the data allows. If you don't have the dynamic range there is not a lot you can do but to stretch less aggressively. I don't believe this Super-L thing is for real and I'm not using it. PI uses, I think, the Lab model internally so stretching would only operate on the luminance component, no matter what you do. As far as I would consider normal dynamic range I only use a 2-step stretching for diffuse objects: MaskedStretch and Arcsinhstretch, in sequence. That is the only thing you need to pull out the details in the shadows and the object's colours. The rest is just masking and curves, as far as I am concerned.

Cheers

Hi Andrea

Thanks again for feedback, your comments make much sense and will keep them in mind for new processing!

CS,
Jure