Jump to content

Banner.jpg.b89429c566825f6ab32bcafbada449c9.jpg

Best approach combining HA+OIII and RGB-Data of Rosette


Recommended Posts

Hi everybody.

I've got tons of data lying around of the Rosette Nebula. I've not started processing as i really can't figure out whats the best approach.

I see 2 ways:

1) Finalizing RGB image and then enhancing Blue+Green with OIII, and enhancing Red with HA (similar to HaRGB just for all 3 channels)

2) Finalizing RGB image, and finalize 'standalone' Bi-Color HA/OIII, and then adding the stars from the RGB to the BiColor image.

Any thoughts?

Kind regards, Graem

Link to comment
Share on other sites

Responding to your PM, I've always done it the first way - with one refinement.

Make three images, starting with Ha applied to red in blend mode lighten and possibly using a hint of Ha as lum as well, being very careful not to overdo it. Call this Ha to red.

Take a copy of Ha to red and apply the OIII to green in blend mode lighten. Call this OIII to green.

Take a second copy of Ha to red and add the OIII to blue in blend mode lighten. Save as OIII to blue.

Now stack up three layers in Ps in this order (though the top two can be reversed.)

Ha to blue

Ha to green

Ha to red

What's nice about this is that by varying the opacity of the top layers you can control the balance of OIII between green and blue while seeing the effect in real time. OIII is really a kind of teal blue if you want to be strict about it.

A note on adding NB in blend mode lighten. Once you have the NB layer in place over the colour layer in BM Lighten you can still manipulate it in Cuves to see if it is doing enough for you. I often pin the bottom of the curve so the NB doesn't affect the background sky but lift it a bit higher up to get more effect in the brighter parts. Some people advise adding the NB in small iterations. Why, though? I pile it all on in one go and then paste that onto the RGB only version. If the NB is overdone I can just reduce the opacity of the NB-modified layer to get it where I want it, flatten and there we are.

I also remind myself that a NB layer for adding to a colour layer is not a standalone image so I can process it differently. I can stretch it above the noise floor, for starters, because the dark parts of the image will be darker than they are in the colour layer so BM Lighten will not apply them. (Cute!) I also know that blending with a colour layer will reduce contrasts so I can pump those contrasts sky high in the NB layer.

Olly

Link to comment
Share on other sites

12 minutes ago, ollypenrice said:

Responding to your PM, I've always done it the first way - with one refinement.

Make three images, starting with Ha applied to red in blend mode lighten and possibly using a hint of Ha as lum as well, being very careful not to overdo it. Call this Ha to red.

Take a copy of Ha to red and apply the OIII to green in blend mode lighten. Call this OIII to green.

Take a second copy of Ha to red and add the OIII to blue in blend mode lighten. Save as OIII to blue.

Now stack up three layers in Ps in this order (though the top two can be reversed.)

Ha to blue

Ha to green

Ha to red

What's nice about this is that by varying the opacity of the top layers you can control the balance of OIII between green and blue while seeing the effect in real time. OIII is really a kind of teal blue if you want to be strict about it.

A note on adding NB in blend mode lighten. Once you have the NB layer in place over the colour layer in BM Lighten you can still manipulate it in Cuves to see if it is doing enough for you. I often pin the bottom of the curve so the NB doesn't affect the background sky but lift it a bit higher up to get more effect in the brighter parts. Some people advise adding the NB in small iterations. Why, though? I pile it all on in one go and then paste that onto the RGB only version. If the NB is overdone I can just reduce the opacity of the NB-modified layer to get it where I want it, flatten and there we are.

I also remind myself that a NB layer for adding to a colour layer is not a standalone image so I can process it differently. I can stretch it above the noise floor, for starters, because the dark parts of the image will be darker than they are in the colour layer so BM Lighten will not apply them. (Cute!) I also know that blending with a colour layer will reduce contrasts so I can pump those contrasts sky high in the NB layer.

Olly

Hi Olly.

Thank you for your thoughts! Make sense.

I have just experimented a little, and have found another third way, that also has some advantages:

- Make a tone map of the bicolor narrowband (starless)

- Add this starless tone map over the RGB (lighten / screen)

- Finalize 'normal' bi-color narrowband image (with narrowband stars)

- Lay the combined tone map & RGB-image over the bi-color narrowband image,  with mode 'color'

This is assuming though that your narrowband stars are in similar size to the RGB stars, what in my case (strangely) is, even if normally narrowband stars are smaller.

Will give 1) a try and come back with my findings.

Kind regards, Graem

Link to comment
Share on other sites

I think it boils down to adding 10% to 90%, and not trying to add 90% to 10%.

I mean that for example with the rosette, 90% of the data i want is in the nebula its self (so in the OIII & HA), and it will be easier to add color to the narrowband stars, than trying to add all that nebula (90%) to the RGB image. I hope i'm making some kind of sense, but i think for the rosette the most efficient 'base line' would be the nebula, and just adding stars (or even as in my previous post, just the colours of the rgb stars)

Not sure yet, but thats my gut feeling currently!

Will update the thread as soon as i have experimented more and found the most efficient way for this target.

Kind regards, Graem

Link to comment
Share on other sites

2 hours ago, ollypenrice said:

Responding to your PM, I've always done it the first way - with one refinement.

Make three images, starting with Ha applied to red in blend mode lighten and possibly using a hint of Ha as lum as well, being very careful not to overdo it. Call this Ha to red.

Take a copy of Ha to red and apply the OIII to green in blend mode lighten. Call this OIII to green.

Take a second copy of Ha to red and add the OIII to blue in blend mode lighten. Save as OIII to blue.

Now stack up three layers in Ps in this order (though the top two can be reversed.)

Ha to blue

Ha to green

Ha to red

What's nice about this is that by varying the opacity of the top layers you can control the balance of OIII between green and blue while seeing the effect in real time. OIII is really a kind of teal blue if you want to be strict about it.

A note on adding NB in blend mode lighten. Once you have the NB layer in place over the colour layer in BM Lighten you can still manipulate it in Cuves to see if it is doing enough for you. I often pin the bottom of the curve so the NB doesn't affect the background sky but lift it a bit higher up to get more effect in the brighter parts. Some people advise adding the NB in small iterations. Why, though? I pile it all on in one go and then paste that onto the RGB only version. If the NB is overdone I can just reduce the opacity of the NB-modified layer to get it where I want it, flatten and there we are.

I also remind myself that a NB layer for adding to a colour layer is not a standalone image so I can process it differently. I can stretch it above the noise floor, for starters, because the dark parts of the image will be darker than they are in the colour layer so BM Lighten will not apply them. (Cute!) I also know that blending with a colour layer will reduce contrasts so I can pump those contrasts sky high in the NB layer.

Olly

I just tried this approach, it works - but i'm confused with one thing:

After you've got these 3 master images, and layered them, playing with opacity, you're also playing with the star color! that can't be right, i must be missunderstanding something in the process?

Kind regards, Graem

Link to comment
Share on other sites

2 hours ago, graemlourens said:

I just tried this approach, it works - but i'm confused with one thing:

After you've got these 3 master images, and layered them, playing with opacity, you're also playing with the star color! that can't be right, i must be missunderstanding something in the process?

Kind regards, Graem

You shouldn't be affecting star colour. It's because your NB stars are too big and/or too bright. You can safely run any number of iterations of Noel's star reduction because, once your NB stars are small/faint enough, they can be artificial lookng and a bit out of shape but Blend Mode Lighten won't apply them to the colour layer. OK, maybe I'm overstating the case but that's more or less how I look at this problem. Get those NB stars down by any means you can!

Olly

Link to comment
Share on other sites

3 minutes ago, ollypenrice said:

You shouldn't be affecting star colour. It's because your NB stars are too big and/or too bright. You can safely run any number of iterations of Noel's star reduction because, once your NB stars are small/faint enough, they can be artificial lookng and a bit out of shape but Blend Mode Lighten won't apply them to the colour layer. OK, maybe I'm overstating the case but that's more or less how I look at this problem. Get those NB stars down by any means you can!

Olly

I think we're missunderstanding each other. I'm aware of that when your adding OIII/HA to the channels, you're keeping the RGB stars, and only adding nebulosity (as NB stars tend to be smaller), but i mean after this step, when you have 'Ha to red', 'OIII to green' etc etc, and you're changing opacity, those are each 3 RGB images enhanced with either OIII or HA, but in this balancing, you're balancing the colors, but also the colors of the stars, or am i missing something?

Kind regards, Graem

Link to comment
Share on other sites

6 minutes ago, graemlourens said:

I think we're missunderstanding each other. I'm aware of that when your adding OIII/HA to the channels, you're keeping the RGB stars, and only adding nebulosity (as NB stars tend to be smaller), but i mean after this step, when you have 'Ha to red', 'OIII to green' etc etc, and you're changing opacity, those are each 3 RGB images enhanced with either OIII or HA, but in this balancing, you're balancing the colors, but also the colors of the stars, or am i missing something?

Kind regards, Graem

Yes, I won't say that you're missing something but that something is not working. In none of the three layers should the addition of the NB to a colour have affected star colour. In that case all three layers should have identical star colour and changing layer opacity should, therefore, have no effect. What must have happened is that the NB has got into the stars. That's what you need to fix.

Olly

Link to comment
Share on other sites

9 hours ago, ollypenrice said:

Yes, I won't say that you're missing something but that something is not working. In none of the three layers should the addition of the NB to a colour have affected star colour. In that case all three layers should have identical star colour and changing layer opacity should, therefore, have no effect. What must have happened is that the NB has got into the stars. That's what you need to fix.

Olly

Olly i get it now, a logic mistake on my part! :) it was a hypothetical question, i did not really see it happening in the data. Obviously you're right, i thought i must be missing something!

I've finished my tone mapping & star stealing approach yesterday (looks promising), and will now still give your way a second try. What i'm missing in your approach is the classical luminance handling the HA & OIII-data (that i got used to). I figure all the contrast enhancing etc is done AFTER these steps on the combined image?

Kind regards, Graem

Link to comment
Share on other sites

2 hours ago, graemlourens said:

Olly i get it now, a logic mistake on my part! :) it was a hypothetical question, i did not really see it happening in the data. Obviously you're right, i thought i must be missing something!

I've finished my tone mapping & star stealing approach yesterday (looks promising), and will now still give your way a second try. What i'm missing in your approach is the classical luminance handling the HA & OIII-data (that i got used to). I figure all the contrast enhancing etc is done AFTER these steps on the combined image?

Kind regards, Graem

I had a spell of doing HaRGB or HaOIIIRGB without luminance and, if a target really is absolutely dominated by emission nebulosity and there is no dusty interest in the background sky, then I'd still do it. However, influenced by Tom I went back to shooting L on most targets. It's hard to keep the stars down but you do get more faint stuff in the background sky.

I would work up the contrasts in the NB before adding it to the colours and then work it up a second time when all is assembled. A little Ha as luminance can also help contrasts but it can wash out colour. Do you know the simple trick in Ps of going to Selective Colour, Reds, and lowering the cyan in red? It emphasizes the deeper reds - ie the Ha signal.

Olly

Link to comment
Share on other sites

I yesterday finally finished experimenting with the third version i mentioned and have gotten to this point (not final, especially something in the red does not let me rest, its not there yet where i would like it!)

@ollypenriceyour way works just as well, but as my fov is strongly dominated by the nebula, its really easier just pulling the star color into the bicolor narrowband image. If there would be a substancial part of normal background & stars, then i'd do it your way, and will try and find another target that allows this.

The eastern and western veil would be perfect targets for your approach i think! I'm empted to try it, as i already have bicolor images there, and just plain white stars.

Here's the effort until now. Comparing a version without star color, and with star color, i'm not sure yet what to prefer, but the blue (purple in my case :p ) stars do give a contrast to the dominantly red ha, what do you think?

591ad9e10c394_NGC2237copy.thumb.jpg.fce9fa1d53924c7ebe0eb6fbc14fd643.jpgNGC2237.thumb.jpg.733ea880cff3e22b0c1009cee7ac7b8c.jpg

Link to comment
Share on other sites

10 minutes ago, ollypenrice said:

It's very 'neboulous' which is, of course good! You should be able to correct the colour in the stars and then I'd prefer them to the white ones.

Olly

I've been trying my **** off yesterday correcting those purple stars to blue ones. 
If you have any hints, i'd amazingly appreciate it.... in PS none of the tools really brought me anywhere (except removing the purple, but then i'm loosing star color in 70% of the stars)

Kind regards,

Link to comment
Share on other sites

graem.thumb.JPG.c9ca53c6a41d1dd165c5c4ccba18b23b.JPG

Working very quickly, and just on a screen grab of your image, I obtained the result above. What I would normally do would be check out the colour index of the 'doubtful' stars and keep that in mind while working on the colour. I didn't do that here, (busy in the robotic shed during the moon.) I just took the view that there are no magenta stars. (There are certainly stars whose blackbody peak lies in the UV but that's a different matter.)

So, take your image and copy layer. Top layer, Noel's Actions (or other way of selecting stars), expand and feather this selection so it just identifies the stars but not what is around them. Erase the star selection from the top. Make bottom layer active. Use Selective Colour on reds, blues, cyans and magentas to get rid of that magenta domination. Also use Hue and Saturation to tweak out the magentas. Flatten and done.

Olly

Link to comment
Share on other sites

On 5/17/2017 at 20:06, ollypenrice said:

 

Working very quickly, and just on a screen grab of your image, I obtained the result above. What I would normally do would be check out the colour index of the 'doubtful' stars and keep that in mind while working on the colour. I didn't do that here, (busy in the robotic shed during the moon.) I just took the view that there are no magenta stars. (There are certainly stars whose blackbody peak lies in the UV but that's a different matter.)

So, take your image and copy layer. Top layer, Noel's Actions (or other way of selecting stars), expand and feather this selection so it just identifies the stars but not what is around them. Erase the star selection from the top. Make bottom layer active. Use Selective Colour on reds, blues, cyans and magentas to get rid of that magenta domination. Also use Hue and Saturation to tweak out the magentas. Flatten and done.

Olly

Thank you very much Olly. I will have a go at this as soon as possible! This is one big issue i still have never resolved with my images, due to the squinting of the ED80. I would be very happy mastering this flaw in post processing.

Kind regards, Graem

Link to comment
Share on other sites

Archived

This topic is now archived and is closed to further replies.

  • Recently Browsing   0 members

    • No registered users viewing this page.
×
×
  • Create New...

Important Information

We have placed cookies on your device to help make this website better. You can adjust your cookie settings, otherwise we'll assume you're okay to continue. By using this site, you agree to our Terms of Use.