Hi all,
I am still learning mono imaging and Pixinsight. Historically, I have struggled with stretching and have had very flat images.
In this particular image, I am not sure the color is correct for the SHO palette.
Please let me know your thoughts. From your experience and knowledge, what am I doing wrong? What steps am I over-doing or may have skipped?
Any advice at all will be greatly appreciated.
Link: https://app.astrobin.com/u/tom62e?i=9kjrjd
Thanks,
Thomas
Just realized I forgot to flip the image to its more common orientation.
Overall I think there’s too much contrast. The colors are garish and the dark parts of the nebula look almost clipped and flat. I suspect that you are over-reacting to the perceived flatness of your stretch results. To really get a handle on stretching I would avoid getting fancy and keep it basic until you get a good feel for it. Avoid things like GHS and stick with a simple histogram transform. Adjust the highlight and black point sliders gradually in steps until you get a well graded image with detail in the highlights, creamy mid-tones and a black point that’s not clipped. You can fine tune that with curves if you need to but avoid large scale changes. I would suggest getting that right before worrying to much about color.
1) The darker areas of the nebula are clipped black.
2) There is too much contrast (part of the reason for the clipped blacks.
3) Too much noise reduction. After noise reduction you should expect to see some remnant noise. That adds to both a sense of sharpness and reality.
Cheers,
Kevin
Thanks guys. I see what you mean. I’m over compensating. I used dark structures enhanced and local histogram equalization. I need to calm them down.
Tommy Mastro · Jul 5, 2026 at 01:33 AM
Thanks guys. I see what you mean. I’m over compensating. I used dark structures enhanced and local histogram equalization. I need to calm them down.
Stretching is the key to this game. I’d learn all of the manual methods before trying those tools. That said I never use those tools! Just a very conservative initial STF/Histogram Transformation stretch followed by some gradual curves adjustments and then shadows and highlights adjustments in photoshop.
Thanks Kevin. I’ve been trying to learn HistogramTransformation for stretching. and MAS.
I accidentally deleted the image from Astrobin attempting to upload a Revision and had to reload it.
The new link in here: https://app.astrobin.com/u/tom62e?i=9kjrjd
Tommy Mastro · Jul 5, 2026, 07:16 AM
I accidentally deleted the image from Astrobin attempting to upload a Revision and had to reload it.
The new link in here: https://app.astrobin.com/u/tom62e?i=9kjrjd
Just wanted to say, in spite of the fair criticism, this is a very nice image, especially the pillars :)
If you improve on the processing as suggested above, it could be amazing :)
Thank you.
I tried lowering the contrast but it seems to not undo properly. I’m going to try reprocessing it from scratch. This way I can compare 1st attempt vis-a-vis second attempt.
Tommy Mastro · Jul 5, 2026, 02:58 PM
Thank you.
I tried lowering the contrast but it seems to not undo properly. I’m going to try reprocessing it from scratch. This way I can compare 1st attempt vis-a-vis second attempt.
I always save intermediate versions of images, named including the last process.
I can easily go back to earlier version without starting scratch.l
I have a small sensor though, doing this can eat storage space.
To me it looks like your gradient extraction took a lot of the nebulosity out, hence the “contrasty” look others have spotted. What’s your workflow for removing gradients? Can you post screenshots of how the data looked like before and after gradient extraction?
TiffsAndAstro · Jul 5, 2026, 03:45 PM
Tommy Mastro · Jul 5, 2026, 02:58 PM
Thank you.
I tried lowering the contrast but it seems to not undo properly. I’m going to try reprocessing it from scratch. This way I can compare 1st attempt vis-a-vis second attempt.
I always save intermediate versions of images, named including the last process.
I can easily go back to earlier version without starting scratch.l
I have a small sensor though, doing this can eat storage space.
That a great idea, thanks!
Mikołaj Wadowski · Jul 5, 2026, 04:02 PM
To me it looks like your gradient extraction took a lot of the nebulosity out, hence the “contrasty” look others have spotted. What’s your workflow for removing gradients? Can you post screenshots of how the data looked like before and after gradient extraction?
📷 Before_Gradient_Removal.jpg
📷 After_Gradient_Removal.jpg
After I do Astrometric Solution and Spectrophotometric Flux Calibration, I use the Multiscale Gradient Correction tool. It’s the best tool I’ve used so far and doesn’t leave any weird artifacts like some of the plug-ins. My seconds favorite is the standard PI Gradient Correction tool.
What do you think? Is it over-doing it?
I think I figured out one boo-boo I made. Using Multiscale Adaptive Stretch I had “Contrast Recovery” checked and set to 100% intensity, when I probably didn’t need it checked at all.
Tommy Mastro · Jul 5, 2026, 06:47 PM
Mikołaj Wadowski · Jul 5, 2026, 04:02 PM
To me it looks like your gradient extraction took a lot of the nebulosity out, hence the “contrasty” look others have spotted. What’s your workflow for removing gradients? Can you post screenshots of how the data looked like before and after gradient extraction?
📷 Before_Gradient_Removal.jpg
What do you think? Is it over-doing it?
Gradient removal isn’t very good and you shouldn’t have much at all, generally is only mildly non-linear (over the image). Graxpert should do it easily enough but before that crop the borders as any drastic change in luminosity there will cause havoc.
Tommy Mastro · Jul 5, 2026, 06:47 PM
Mikołaj Wadowski · Jul 5, 2026, 04:02 PM
To me it looks like your gradient extraction took a lot of the nebulosity out, hence the “contrasty” look others have spotted. What’s your workflow for removing gradients? Can you post screenshots of how the data looked like before and after gradient extraction?
📷 Before_Gradient_Removal.jpg
What do you think? Is it over-doing it?
MGC is okay for widefield images, but for your pixel scale I don’t think it would work very well. Additionally here, your scale factors are definitely off judging by how large scale Ha and Oiii layers got subtracted, and by how they got added to the Sii image. I recommend either asking someone for a widefield reference or shooting one yourself, and then using it to correct the gradient.
Regarldess, these do not look like the same files you used to create the final image, as the final version does not have this Sii “fog”,
Personally this structure over/under subtraction is the biggest flaw in your image. You can take a look at NSNS survey data (effectively gradient-free) and compare how Sii, Ha, and Oiii actually look like vs how they appear in your image:
Notice how the areas circles in green (parts with real narrowband signal) end up darker than the background, highlighted in red:
The background in general appears sort of inverted, brighter than it should be. Both are rather objective technical flaws.
They are usually signs of poor background extraction. They could have also appeared after applying LHE or dark structure enhancement, both of which I recommend applying very lightly or avoiding all together as they can often “cook” the image. Without knowing your entire workflow it’s difficult to pinpoint the exact moment these artifacts appeared.
Since the raw data is pretty decent, I’m going to try processing it again from scratch trying to be mindful of what everyone here has said.
Miko/Andrea - Thanks for the feedback. I’m going to try and be gentler with the background extraction and I will eliminate LHE and DSE. Unfortunately, Graxpert never works for me. Even if I crop first, it always leaves weird artifacts and introduces significant noise. Gradient Correction seems to work smoothly, even if it isn’t the best. Perhaps it will be less aggressive than MGC.
I was just practicing with Multiscale Adaptive Stretch. I think I have the hang of it now. I just need to develop a better eye. This may have contributed to the bright background and the contrast recovery may have over-cooked the dark spots (that I assumed were dark nebula) even before DSE cooked it.
I suspect its all of these things together that are contributing to my issues. Practice makes perfect. Thanks everyone!
Tommy
I have the same idea that contrast is too much. If you are using LHE (I usually use this one for mid-large scale contrast), be a little more gentle!
Ha! I figured out why Graxpert hates me.
For one thing, I had a much older version. Also, I never realized there was a process for it. I always ran it from Scripts (not sure if that makes a difference).
Anyway, now with the latest version, it works like a charm!
Thanks Andrea et al.!