[RCC] Rosette Nebula in SHO/Foraxx Requests for constructive critique · Jan Ossowski · ... · 7 · 557 · 0

janekosa 1.51
...
· 
·  1 like
·  Share link
Hi all,
I am asking for honest opinions and suggestions about my rosette image. 
The acquisition details can be found in the photo. This is my third narrowband attempt and second picture processed in pixinsight, I'm still on trial

I have always struggled to get the background/contrast just right. 
​​​​​​This image has 2 revisions and they were processed completely independently. 

On the first attempt i made a shortcut of extracting the background by substraction. It was a very crude pixel math process, something along (picture - 0.8*background +0.05).
This resulted in nice contrast in the dim parts of the nebulosity, but it bothered me and after a while of looking at it despite getting mostly positive feedback, it looked artificial to me (well, it is). That's mostly why I decided to make a second attempt.

On the second try I extracted the background by division, and then used GeneralizedHyperbolicStretch (many subtle iterations) to get the contrast right. I think this looks much more natural and I definitely like it more personally, but I am of course not objective. Again, I got mostly positive feedback and most people say this version is better, but apparently many people just like the colors more. I also got feedback that the luminosity part of the first picture is actually better, the background is too dark and there is not enough contrast. 

I would like to hear more opinions and suggestions on what process to use. 

I am obviously also open to all feedback even if not related ​to the aspect I mentioned

Here is the photo in question:
https://www.astrobin.com/i3pr2c/B/

Lastly, here is a shortcut of my process:
- wbpp (drizzle 2x)
- photometric mosaic (per channel)
- blurX
- resample
- starX
- noiseX
- background extraction by division (per channel)
- statistical stretch (as preliminary step)
- foraxx script
- another round of background extraction by division (got very blue background because of uneven stretch of channels)
- multiple iterations of GHS as mentioned
- curves adjustments with convoluted color masks

I may have missed something but that captures the main points at least. 

Thanks in advance and sorry for long post 
​​​​​
Edited ...
Like
Rjhat3 0.00
...
· 
·  Share link
I'm no expert by any means, but the image is really good. Stars are round and not over exposed, the detail and contrast look good.
The only thing I can say comes down to opinion and taste.
I personally think the colors are slightly over saturated. BUT, that's my opinion. Based on the many, many narrowband images on astrobin, some probably would prefer the saturation.
The background in this image is a bit clipped and it appears that you lost some detail along the outer perimeter of the ring.
I am leaning more towards the first image. The background is not as black and I can see those details that are missing in the newer version.

Your processing skills are good, especially if you are relatively new! PixInsight is a huge undertaking to completely comprehend and is (in my opinion) one of the more difficult parts of AP to learn, assuming your equipment choices were decent. I am still learning myself. I suspect everyone here learns something every time they process an image.
Over all, great job! Very nice image!
Edited ...
Like
Alexn 12.25
...
· 
·  Share link
Looks lovely to me!
Like
WhooptieDo 10.40
...
· 
·  Share link
I actually like your production here, especially if pixinsight is still new to you.   I think your final image is very respectable.  You managed all the detail, you did NOT overdo it with BlurX which is something I see all too common these days.   The noise was managed to a decent level and the background seems fairly neutral.   The colors are a tad bold for me, but that's not a fault, just a difference in taste.

Since you are asking for critique, I will say you almost overdid it with BlurX.   This becomes a bit evident with a few of the small bok globules that have unnaturally sharp edges.   The rest of the image however doesn't seem to exhibit this.    The process of drizzle, blurx, then downsample is a rather powerful little trick some of us do, and I think it worked rather well for you here.

The other thing that stands out to me is your stars seem to have lost their intensity.  I'm not sure if you added them back to your image properly.  Size, shape, color all seem nice, but they're just soft, and maybe a tad dimmer than they should be.   Try Charles Hagen's star addition guide here.  It's a generally accepted method that most of us use.

On a side note, unless you are having flat issues, uncorrected vingetting, etc in your image.   There is no reason to use 'division' method with background extraction.

The fact that you're using convoluted masks already in some of your early pix productions kind of impresses me.  I don't see people pick up on this very early.    I recommend trying MMT on large scales to create better masks.   convolution just isn't quite my forte, but it gets the same job done.

Also, as I read this, I just saw your first/brighter revision, and I will say I'm not as big of a fan of that one.  The background feels desaturated, like extra effort was put in to make sure it was neutral, and it doesn't feel natural.     The bolder revision is a much better image.
Like
janekosa 1.51
Topic starter
...
· 
·  Share link
Rodney Hatchett:
Your processing skills are good, especially if you are relatively new! PixInsight is a huge undertaking to completely comprehend and is (in my opinion) one of the more difficult parts of AP to learn, assuming your equipment choices were decent. I am still learning myself. I suspect everyone here learns something every time they process an image.
Over all, great job! Very nice image!

Alex Nicholas:
Looks lovely to me!

Brian Puhl:
The fact that you're using convoluted masks already in some of your early pix productions kind of impresses me.

Thank you, but honestly it's not much of an achievement. Pix is new software for me, and I'm still amazed at some of the capabilities and getting used to the workflow (vastly different from layer based software such as photoshop) but in essence it's just processing software. It's not like there are no color masks in photoshop, it's not like the overall image process is completely different.
Rodney Hatchett:
I personally think the colors are slightly over saturated.

Brian Puhl:
The colors are a tad bold for me, but that's not a fault, just a difference in taste.

Yeah I'm probably guilty as charged here. I just don't find the typical brown + brown + a bit of blueish brown SHO palette (i know, i know, i'm just making a point here) all that exciting. Maybe it comes from too much instagram and flashy everything these days, maybe it's something that will come with experience. I do my best to keep the colors within acceptance threshold for everyone, but what can I say.. I like them saturated. 
Brian Puhl:
The noise was managed to a decent level and the background seems fairly neutral.

Brian Puhl:
not as big of a fan of that one.  The background feels desaturated, like extra effort was put in to make sure it was neutral, and it doesn't feel natural.

I am very glad you think that. As I mentioned this was my very goal and my motivation respectively for the second process. Seems like it was at least a partial success then
Brian Puhl:
The other thing that stands out to me is your stars seem to have lost their intensity.  I'm not sure if you added them back to your image properly.  Size, shape, color all seem nice, but they're just soft, and maybe a tad dimmer than they should be.   Try Charles Hagen's star addition guide here.  It's a generally accepted method that most of us use.

Yeah, that's something that is (like background) an art of balance between making the stars unnaturaly dim and making them so bright that they draw attention away from the image. Seems like I failed to find this perfect balance this time. The gist of my process was as described. Thanks for this feedback though, that's something I'll pay more attention to next time.

Brian Puhl:
On a side note, unless you are having flat issues, uncorrected vingetting, etc in your image.   There is no reason to use 'division' method with background extraction.

What would you say is the right way of background extraction for such cases? Let's say there is no gradient or vignetting whatsoever, but there are large differences between channels in background levels. Am I supposed to just non-linearily stretch individual images until the backgrounds are similar? Is there a tool that makes it easier? As I said, this is one of the things I feel I struggle with the most. The division almost felt like magic in terms of how easy it was and how good results it produced.
Like
WhooptieDo 10.40
...
· 
·  1 like
·  Share link
Jan Ossowski:
Brian Puhl:
The other thing that stands out to me is your stars seem to have lost their intensity.  I'm not sure if you added them back to your image properly.  Size, shape, color all seem nice, but they're just soft, and maybe a tad dimmer than they should be.   Try Charles Hagen's star addition guide here.  It's a generally accepted method that most of us use.

Yeah, that's something that is (like background) an art of balance between making the stars unnaturaly dim and making them so bright that they draw attention away from the image. Seems like I failed to find this perfect balance this time. The gist of my process was as described. Thanks for this feedback though, that's something I'll pay more attention to next time.

Brian Puhl:
On a side note, unless you are having flat issues, uncorrected vingetting, etc in your image.   There is no reason to use 'division' method with background extraction.

What would you say is the right way of background extraction for such cases? Let's say there is no gradient or vignetting whatsoever, but there are large differences between channels in background levels. Am I supposed to just non-linearily stretch individual images until the backgrounds are similar? Is there a tool that makes it easier? As I said, this is one of the things I feel I struggle with the most. The division almost felt like magic in terms of how easy it was and how good results it produced.



Just to add, star processing isn't difficult, and shouldn't require a balance.   Your white point is off really, probably because they needed to be re-linearized when put back into the image.  See my prior reference for that.

If there are no gradients, and no vignetting, then there is no need for background extraction.  This is not uncommon with tight narrowband filters.     You generally don't individually stretch the images, you combine them linearly into something like the SHO palette, background neutralize, then color calibrate.    Now things like Foraxx's palette, I believe are non-linear formulae, unfortunately they have to be stretched ahead of time.    For situations like that, I pretty much just do a boosted STF on my combined image, which will expose the darker areas for their true colors.   Open up histo transformation and either align the black points, or manually tweak it until it visually looks neutral.
Like
TiffsAndAstro 1.81
...
· 
·  1 like
·  Share link
Jan Ossowski:
Hi all,
I am asking for honest opinions and suggestions about my rosette image. 
The acquisition details can be found in the photo. This is my third narrowband attempt and second picture processed in pixinsight, I'm still on trial

I have always struggled to get the background/contrast just right. 
​​​​​​This image has 2 revisions and they were processed completely independently. 

On the first attempt i made a shortcut of extracting the background by substraction. It was a very crude pixel math process, something along (picture - 0.8*background +0.05).
This resulted in nice contrast in the dim parts of the nebulosity, but it bothered me and after a while of looking at it despite getting mostly positive feedback, it looked artificial to me (well, it is). That's mostly why I decided to make a second attempt.

On the second try I extracted the background by division, and then used GeneralizedHyperbolicStretch (many subtle iterations) to get the contrast right. I think this looks much more natural and I definitely like it more personally, but I am of course not objective. Again, I got mostly positive feedback and most people say this version is better, but apparently many people just like the colors more. I also got feedback that the luminosity part of the first picture is actually better, the background is too dark and there is not enough contrast. 

I would like to hear more opinions and suggestions on what process to use. 

I am obviously also open to all feedback even if not related ​to the aspect I mentioned

Here is the photo in question:
https://www.astrobin.com/i3pr2c/B/

Lastly, here is a shortcut of my process:
- wbpp (drizzle 2x)
- photometric mosaic (per channel)
- blurX
- resample
- starX
- noiseX
- background extraction by division (per channel)
- statistical stretch (as preliminary step)
- foraxx script
- another round of background extraction by division (got very blue background because of uneven stretch of channels)
- multiple iterations of GHS as mentioned
- curves adjustments with convoluted color masks

I may have missed something but that captures the main points at least. 

Thanks in advance and sorry for long post 
​​​​​


You have been found guilty of taking and processing an amazing image by a jury of your peers.
Do you have any last requests?

Like
janekosa 1.51
Topic starter
...
· 
·  Share link
Thank you all for your kind words and (especially) all the advice. Hopefully I can post another RCC at some point. Right now it doesn't look like I'm gonna get a clear night for a long while though, so I have time to work on my processing ;)
Like
 
Register or login to create to post a reply.