M31 HaLRGB

[Rodd]

2 hours ago, wimvb said:

A lot of careful processing,

A second--more careful look and man--great details.  Really nice.  Highr esolution and everything looks tremendous--except that yellow.  But zoomed in it does not bother me.  Now THATS the core of M31!

Rodd

[Rodd]

 

 

2 hours ago, wimvb said:

There will be a version 2

It is growing on me more.  How the heck do you do that.  That CANT be the same data !  Zoomed in is INCREDIBLE!  I knew there was a reason I thought mine looked like a cartoon....just needed a reality check.  Well done.

Rodd

[Laurin Dave]

Hi Rodd

Fantastic data.. Here's my go...  I'll never get sick of M31  ..  still think that there's more halo to be found as it looks darker near the galaxy than in the corners.

Thanks for posting it.  

Dave 

 

[Rodd]

39 minutes ago, Laurin Dave said:

 still think that there's more halo to be found as it looks darker near the galaxy than in the corners.

Dave--Nice.  You may be right.  Wim found a slight stacking error I missed for cropping.  I have since redone the stacks--being more careful with the crop and the DBE.  It has resulted in a smoother, more consistent background--but I think it removed the light toward the corners as opposed to lifting the halo.  Remember--these were only 60 sec subs.  Not like Ollys 20 min lum composition, in which he really captures the extended glow.  maybe you are looking for what you know to be there but isn't due to the data just not reaching that depth.   

I have rethought my entire processing approach--again.  It is painfully obvious to me that my image just does not stand up to....to...well to much of anything.  Take a good look at full resolution viewing of Wims image.  It's hard to believe its the same data.  I don't have access to a high-resolution screen at present (which really makes a difference), so yours doesn't look as good as it probably is to me. But it does not have any elements of the cartoon.  Your image is very pleasing--real looking.  Mine belongs in a Disney poster--maybe Aladdin's venture through the cosmos.  I don't know.  I am reprocessing the data.  Maybe I should just stick to data collection.

Rodd

[Rodd]

1 hour ago, Laurin Dave said:

. Here's my go

Just looked on my iPhone (which are great screens for this) and the image looks fantastic.

Rodd

[kirkster501]

Great pictures.

The point to remember is what *does* M31 look like?  It is a ghostly black and white smudge in the sky and there is no single colour standard.  The galaxy would never actually look like these pictures if you were 100000 light years above it. Anymore than our own Milky Way does and we are inside it.

Artistic licence is what applies and if the picture is good to your eyes then you have hit the bullseye and that's all that matters.

[Laurin Dave]

22 minutes ago, Rodd said:

Just looked on my iPhone (which are great screens for this) and the image looks fantastic.

Rodd

Thanks Rodd...   I generally use a 21.5" Imac for processing, it and my iPhone and iPad knock the socks of my Toshiba laptop and HP desktop..  Out of interest I used Photometric colour calibration and saturation boost in curves, although whether this means the colours are anyway accurate I have no idea!

Dave

[Rodd]

5 minutes ago, kirkster501 said:

Great pictures.

The point to remember is what *does* M31 look like?  It is a ghostly black and white smudge in the sky and there is no single colour standard.  The galaxy would never actually look like these pictures if you were 100000 light years above it. Anymore than our own Milky Way does and we are inside it.

Artistic licence is what applies and if the picture is good to your eyes then you have hit the bullseye and that's all that matters.

True enough--the universe is not a colorful place.  However-- the structure is there--we just can't see it due to the dynamic range and dimness of the emissions.  But the structure is there.  That's why when data is processed well, the structures look real--like they have depth.  its photographic accuracy as far as the boundary of the structure is concerned.   Kind of like the rim of a planet--the features within the disc may not look like the wonderfully processed images of planets--but the edge of the disc looks the same.  the same can be said for some Bok globules--or in this case, some of the dust lanes in M31.  The task of the image/processor is to modify the dynamic range so that the features CAN BE distinguished, without altering any other aspect of the object.  Palette is in great part artistic license, as you say.  But not wholly, especially in broad-band.  Be that as it may, even if you are entirely correct and it is ALL artistic license, one must determine before hand what type of art they wish to create.  In my case, I am producing Picasso's (his later stuff--not his early drawings), when I want to be creating an (insert name of a realism artist you are familiar with.  Robert Zapalardi comes to mind--but you probably don't know him.  So choose a landscape master of photographic quality)

All I know is I have gone too far down this dead end alley.  It is a secret entrance to the world of Roger Rabbit, or to the infamous rabbit hole.  

Rodd

[Rodd]

10 minutes ago, Laurin Dave said:

Thanks Rodd...   I generally use a 21.5" Imac for processing, it and my iPhone and iPad knock the socks of my Toshiba laptop and HP desktop..  Out of interest I used Photometric colour calibration and saturation boost in curves, although whether this means the colours are anyway accurate I have no idea!

Dave

M31 is notoriously difficult regarding palette.  Yours looks good to me.  I have been told it tends more toward blue than red/green.  your Ha looks great.  If you spent many hours tweaking here and there, I am sure you will come up with a very finely detailed, crisp masterpiece.  At present, you M31 looks like I imagine it would look using the Keck telescope and a Televue Ethos and a bino viewer.    And this on a cheap screen.  Its a great image--not too far pushed in palette or sharpening or noise control or star control.  Well done.

Rodd

[coatesg]

Just started to take a look - the data is very nice, but I think I agree with the comments on the DBE - it's possibly overcooked to the galaxy as there appears to be a slightly dark halo around it - not much, but it makes grabbing the faint halo very hard indeed. I get this occasionally as well if I have points covering a star, but it could be the points err too close to the galaxy itself. (It could also be flats overcorrecting I suppose).

The raw stacks would be interesting to see the difference.

[Laurin Dave]

1 hour ago, Rodd said:

M31 is notoriously difficult regarding palette.  Yours looks good to me.  I have been told it tends more toward blue than red/green.  your Ha looks great.  If you spent many hours tweaking here and there, I am sure you will come up with a very finely detailed, crisp masterpiece.  At present, you M31 looks like I imagine it would look using the Keck telescope and a Televue Ethos and a bino viewer.    And this on a cheap screen.  Its a great image--not too far pushed in palette or sharpening or noise control or star control.  Well done.

Rodd

It’s the data Rodd...  which is yours, good data, and lots of it, makes all the difference 

Dave

[wimvb]

This is how I would approach the flatness problem in @Rodd's image. All samples in DBE are evenly distributed around the main target, with none of them interfering with the faint outer halo. Btw, there's no need to avoid the small stars in samples. The contribution of each pixel in the calculation of the background model is represented by the intensity in the sample square. The brighter a pixel here, the more it will contribute. Black areas don't contribute at all. I just make sure that the black areas are kept to a minimum, and that there are no bright halos around them. The Wx numbers to the left of the sample are the average of all the sample pixels. I keep this above 0.75 at (almost) all times.

Also note that I use normalisation in image correction. This means that the average background is preserved. I never use DBE for background neutralisation.

So far, this method has served me well, so I stick with it. (If it ain't broke, ...)

[gorann]

Rodd, I have not had time to start a PS approach to your data yet but the week end is getting closer, and it is likely to be yet another cloudy one. Any chance you could post you non-DBE treated files? I do have plenty of other non-AP things I really should do so no pressure....

[wimvb]

V 1.5

A little more care taken to keep colour in the core, while also keeping detail near the core. The halos around the stars are still a problem. I tried star reduction on the blue channel only, but this was apparently not enough. The Ha regions are a little bolder in this version.

The large version could take a while to download.

The core in L was slightly blown out. I created a new L master by combining the R, G, and B masters. I then used HDR Composition to repair the original L.

For V 2 I will probably do a complete rework. The weekend will be clouded out anyway.

[Rodd]

9 hours ago, Laurin Dave said:

 

 

It’s the data Rodd..

 

1 hour ago, wimvb said:

The weekend will be clouded out anyway.

 

3 hours ago, gorann said:

Any chance you could post you non-DBE treated files

Well--I finally was able to apply some of the methods that I was able to pick up from Wim's and Dave's image.  I deconvolved the linear lum--but like Wim--I integrated the RGB and Lum stacks using the red channel as reference.  It fixed the core and like Wim I compressed it.  Its amazing to see structure right into the heart of the monster!  But there is a twist--while the details were amazing and the stars looked great--the palette was just not to my liking.  It looked very much like Wim's.  No offense intended, but I am not partial to the yellow.  While zoomed in it looks great--but fro normal viewing I liked my original palette better.  So......I inserted the lu from my new version into the second image I posted in this post.  It preserved the details AND the palette (more or less).  Its funny--but these image--this one and Dave's and Wims look great zoomed in, while they look a bit underwhelming at normal viewing.  Maybe its because there is so much data in such a small space that the eye can not perceive it.  So I posted a crop to to show you guys that I have indeed learned a great deal from you.  While I am not 100% satisfied (It was a long and messy process--and I should be able to render an image in and of itself without resorting to a luminance transfer)--but I at least can retire in peace now, knowing that the heavens will last one more day!!!

Thank you very much 

Rodd

Edit--Stars are a bit pale--working on that now

Goran--I will post the subs--tomorrow.  MUST SLEEP!

 

 

[Rodd]

Here's the image with a bit more saturated stars.  Better?

[don4l]

This has been an absolutely brilliant thread.

That last close up of the core is the best that I have ever seen.

 

 

[Rodd]

13 minutes ago, don4l said:

This has been an absolutely brilliant thread.

That last close up of the core is the best that I have ever seen.

 

 

Thanks Don.  

Rodd

[Rodd]

Here is the image that I used the luminance from.  I think its ok--but the palette is not to my liking.  The details are nice though--not over sharpened like my first!  This is the finished product of this process prior to extracting the lum and inserting it in the previous image for its palette.  It is too yellow--but it is growing on me.

Rodd

 

[gorann]

Rodd, on my job computer the sky color is better in the first one. It looks lilac/blue in the second one.

[Rodd]

33 minutes ago, gorann said:

Rodd, on my job computer the sky color is better in the first one. It looks lilac/blue in the second one.

Thanks Goran--I removed that one--the one I left was the next step after where I neutralized the background, so it was really the final image--the source of the lum.

Rodd

[coatesg]

15 hours ago, wimvb said:

This is how I would approach the flatness problem in @Rodd's image. All samples in DBE are evenly distributed around the main target, with none of them interfering with the faint outer halo. Btw, there's no need to avoid the small stars in samples. The contribution of each pixel in the calculation of the background model is represented by the intensity in the sample square. The brighter a pixel here, the more it will contribute. Black areas don't contribute at all. I just make sure that the black areas are kept to a minimum, and that there are no bright halos around them. The Wx numbers to the left of the sample are the average of all the sample pixels. I keep this above 0.75 at (almost) all times.

Also note that I use normalisation in image correction. This means that the average background is preserved. I never use DBE for background neutralisation.

 

That's how I would approach it as well, but I think that if run instead of the prior DBE, it'd save a step ?

The star issue is interesting - the stars themselves aren't the issue - it's the halos round them! Also, if you have a much higher gradient to deal with, and you need to increase the tolerance to deal with it, you can land up with more of the halo and background stars being dragging into the sample. Depends a lot on the situation I find.

[souls33k3r]

An absolute master class. This thread has plethora of information and an absolute gold mine. @Rodd, really love your posts mate. Gives us mere mortals plenty to learn from yourself, @wimvb and others.

 

[DaveS]

May be late to the party, but those halos look like the effects I've seen during DDP (Using AA5) while I'm getting the values right.

[Rodd]

14 minutes ago, DaveS said:

May be late to the party, but those halos look like the effects I've seen during DDP (Using AA5) while I'm getting the values right.

Didn't use DDP.  I think the halos are the effect of shooting the blue filter during bad conditions.  Also--I think, for the most part, they are due to the luminance and the RGB having different star profiles,  When I inserted the lum the stars were smaller, resulting in the halos.   The stars with the FSQ at F3(.6x reducer) are not great anyway.  I have learned to live with them

Rodd