Did I overstretch this image?

Topic starter

Jun 27, 2022

I posted an image of the Eastern Veil Nebula (NGC 6992) a few days ago:

https://www.astrobin.com/full/ymol78/0/

I'm still kinda new at the imaging thing but I've gotten better in the past few years. One key decision was using a modded DSLR (Nikon D5300 and Nikon D5500).

I wanted the red to come through with the mod, and by God, does it ever. So much so that it looks like there's TOO much red; the H-alpha is simply overwhelming the O-III.

The result in the link is after I did the following in PXI--

1. Crop and ABE
2. Linear Fit, set blue as the reference.
3. Color mask on the blue channel; added a bit of green to make it more aquamarine/turquoise for the O-III.

That's after 2-1/2 hours. I know I need more, but this part of the country (NE Ohio) isn't conducive to 20-hr integrations. That being said, it still needs more. I tried to bring out more given limited imaging time but I think I hit that point where I might have overdone it.

Looking at the image, is there anything further I could have done to balance out the colors? And did I overstretch it?

Clear Skies,
Phil

Helpful Engaging

Observatório Astrográfico do Boqueirão

Jun 27, 2022

No.. in terms of data.. there's plenty enough to work with and you did the stretch as much as it could go in my point of view

andrea tasselli

Frequent contributor

Jun 27, 2022 · Edited: Jun 27, 2022

Your colour balance isn't the best to bring out the cyan of the OIII. And I wouldn't do what you did.

John Hayes

Frequent contributor

Jun 27, 2022

It doesn’t look over-stretched. It looks over-exposed, which is not to say that you need less exposure time. It’s just too bright. It sounds like you are using PI so you might try using the Curves tool to tone it down a bit. You might also try a gentle application of masked LocalHistogramEqualization. Finally, I’d dial back the star field. It’s too overwhelming. Try applying MorphologicalTransformation to reduce the size and brightness of the stars. Then gently adjust the saturation using the Curves tool. Just be sure to mask the background to keep it neutral.

John

Well Written Helpful Insightful

Phil Creed

Topic starter

Jun 27, 2022

andrea tasselli:
Your colour balance isn't the best to bring out the cyan of the OIII. And I wouldn't do what you did.

Duly noted. As a beginner, I figured there's a better way of balancing out the colors and drawing out the O-III, hence the inquiry. How should I have done this considering this was not shot through a narrowband filter? Had it been a narrowband filter, I'd have at least a few ideas on how to do the color balance, but not in this situation.

I mean, that O-III data IS there. How should I have weighed the channels to bring it out more?

Clear Skies,
Phil

Well Written Respectful Engaging

Phil Creed

Topic starter

Jun 27, 2022

John Hayes:
It doesn’t look over-stretched. It looks over-exposed, which is not to say that you need less exposure time. It’s just too bright. It sounds like you are using PI so you might try using the Curves tool to tone it down a bit. You might also try a gentle application of masked LocalHistogramEqualization. Finally, I’d dial back the star field. It’s too overwhelming. Try applying MorphologicalTransformation to reduce the size and brightness of the stars. Then gently adjust the saturation using the Curves tool. Just be sure to mask the background to keep it neutral.

John

John,

I used the EZ Star Reduction in PI and set it to 8 iterations of morphological transformation. But I did that towards the end.

Should I have done that in the linear stage? And should I have done a few rounds of EZ Star Reduction?

Clear Skies,
Phil

Well Written Engaging

John Hayes

Frequent contributor

Jun 27, 2022

Phil,
That's interesting. I usually reduce star size well after the stretch and near the end. How big were the stars to start with??

John

Kevin Morefield

IOTD Staff

Jun 27, 2022

John Hayes:
Phil,
That's interesting. I usually reduce star size well after the stretch and near the end. How big were the stars to start with??

John

Same here - I do star reduction at the end. But I do it in Photoshop for the most part.

In PS it is the Minimum filter that I apply via a donut shaped selection. That is a selection of the brightest maybe two-thirds of the stars MINUS a selection of the saturated or near saturated star cores. After the Minimum I typically add back just a touch of noise to better match the surrounding background.

For the star selections I use the Select > Color Range > Highlights tool.

Helpful

andrea tasselli

Frequent contributor

Jun 27, 2022

Phil Creed:
Duly noted. As a beginner, I figured there's a better way of balancing out the colors and drawing out the O-III, hence the inquiry. How should I have done this considering this was not shot through a narrowband filter? Had it been a narrowband filter, I'd have at least a few ideas on how to do the color balance, but not in this situation.

I mean, that O-III data IS there. How should I have weighed the channels to bring it out more?

Clear Skies,
Phil

My motto is "keep it simple". So, once you have the stacked RGB image try using PhotometricColourCalibration. Once calibrated you're already there unless you didn't filter out the near-IR. Did you?

Secondly, I'd consider morphological transformations a last resort kind of operation. If needs be done it needs to be done at the very end of the image creation pipeline.

Helpful Concise

Phil Creed

Topic starter

Jun 28, 2022

How about this?

I gave it a few rounds of saturation adjustments on the stars and two rounds of EZ Star Reduction. In the linear stage, I did a curves adjustment on the colors and it was MUCH touchier to changing than when the image is stretched but I think it looks better.

Thoughts?

Clear Skies,
Phil

John Hayes

Frequent contributor

Jun 28, 2022

That definitely looks better! The colors are more delicate–and not blown out. And…you've done a nice Jobe of de-emphasizing the star field. I think that you are definitely on the right track.

John

Supportive

Doug Azwell

Jun 28, 2022 · Edited: Jun 29, 2022

I used a different approach. You have tons of data and I used it on the JPG version of the pic in MaximDL. I used a bit of color saturation and then dropped the levels a bit followed by a bit of remove background color. To me it makes the color pop.

Also notice that the Hydrogen alpha areas extend well beyond the wisps of the filamentary nebulosity, so it makes it very hard to find an area devoid of a glow to do a remove background color routine. I had to look around a bit to find a good spot.

It's a great pic worthy of making into a desktop background or a Fracture picture. Awesome shot.

Doug Azwell

Helpful Respectful Engaging Supportive

Seymore Stars

Jun 28, 2022

In PI Image>Extract>Split RGB Channels will produce Greyscale images for each channel. Rebalancing them without ruining the stars is the tricky part.

Starnet removes the stars. They can be recombined with weights applied in pixelMath.

Phil Creed

Topic starter

Jun 28, 2022

Seymore Stars:
In PI Image>Extract>Split RGB Channels will produce Greyscale images for each channel. Rebalancing them without ruining the stars is the tricky part.

Starnet removes the stars. They can be recombined with weights applied in pixelMath.

So if you only took 70% of the red channel, did you still leave the G and B at 1.0X? I know how to do luminance masking to prevent the background from color shifting; it's the colors of the starless nebula that's tough. The .jpg I posted in this thread (10th post) I got the colors strictly by a curves adjustment in L*a*b, but that was VERRRRRY touchy in the linear stage. I mean, just the SLIGHTEST adjustment got me from my original color balance to the one in post #10.

I can keep the star colors unchanged through Starnet++. It's not perfect and at time will grab portions of the intended DSO. Version 2 of Starnet++ seems better in that regard, though.

Clear Skies,
Phil

Helpful Engaging

Seymore Stars

Jun 29, 2022

Yes G and B were kept at 1.

Let me know how it goes.

CS
Steve

Was the linked image overstretched?