Pacman : PixelMath vs. Annie's Astro Action

[Adreneline]

Decided to continue the comparision by using both PixelMath and PS Annie's Astro Action to produce the HOO Bi-Colour image.

I think I prefer Annie over PixelMath!

Annie:

PixelMath:

Thanks for looking.

Adrian

[Xiga]

18 minutes ago, Adreneline said:

Decided to continue the comparision by using both PixelMath and PS Annie's Astro Action to produce the HOO Bi-Colour image.

I think I prefer Annie over PixelMath!

Annie:

PixelMath:

Thanks for looking.

Adrian

Looks good Adrian!

For me though, the Pixel Math one is the clear winner. It has a much more neutral background and the star colours are also better. 

It's not really a fair comparison though. It looks to me like Pixelmath does a great job of getting to the end point on it's own, whereas Annie's Action really needs a Levels and Curves adjustment to fix the colour balance and brightness to bring it more into line with the Pixelmath version. 

Another option is Noel Carboni's 'Synthesize Green Channel' action in PS. I regularly use it, and find it does a great job too. 

[Adreneline]

6 minutes ago, Xiga said:

It's not really a fair comparison though. It looks to me like Pixelmath does a great job of getting to the end point on it's own, whereas Annie's Action really needs a Levels and Curves adjustment to fix the colour balance and brightness to bring it more into line with the Pixelmath version. 

Another option is Noel Carboni's 'Synthesize Green Channel' action in PS. I regularly use it, and find it does a great job too. 

Thank you - that's exactly the sort of feedback and advice I was hoping for. I have the Carboni Actions - I'll give it a go!

Thanks again.

Adrian

[AngryDonkey]

I have never used Annie's Astro Action and can't drawn any comparisons but I've worked a lot with PixelMath recently and have to say that although this seems a pretty simple tool at first, when you get deeper into it, you realise how powerful it can be (If you know how to mathematically express what you are trying to achieve).

For example I didn't realise until recently that I could add conditional statements to the formulas in an Excel like manner. This opens up a lot of possibilities when combining narrowband data. Here is an interesting example I've lifted off Reddit:

Red: iif(ha > .15, ha, (ha*.8)+(oiii*.2))
Green: iif(ha > 0.5, 1-(1-oiii)*(1-(ha-0.5)), oiii *(ha+0.5))
Blue: iif(oiii > .1, oiii, (ha*.3)+(oiii*.2))

It seemed to work well with my HOO data.

[Adreneline]

3 hours ago, AngryDonkey said:

Here is an interesting example I've lifted off Reddit

Thanks Mike. I will give it a go and see what result it produces on my Pacman. I understand there are no doubt limitless possibilities to exoeriment with colour combinations in PixelMath. Annie's Actions require you to make adjustment to Levels in PS as part of the Action and so again, limitless possibilities.

Thanks for looking and for the advice on PI.

Adrian

[wornish]

13 hours ago, Adreneline said:

Decided to continue the comparision by using both PixelMath and PS Annie's Astro Action to produce the HOO Bi-Colour image.

I think I prefer Annie over PixelMath!

 

Thanks for looking.

Adrian

Thanks for the example pixelmath info.  I have not tried it using mathematical formulas for combining channels so far and this will be a next step in my learning curve.

[Adreneline]

4 hours ago, wornish said:

Thanks for the example pixelmath info.

Thanks are due to @AngryDonkey for the PixelMath example.

My approach was much less sophisticated and as presented on the LightVortexAstronomy website, namely:

• R : Ha ; G : OIII ; B : OIII
• R : Ha ; G : (0.4xHa)+(0.6xOIII) ; B : OIII [playing with the 0.4 and 0.6 produce different results)

Good luck!

Adrian

[ollypenrice]

The background sky is a considerably more neutral colour in the Pixelmath image. I might give SCNR green a spin on both of them, too. In the Pixelmath version there's a hint of green in the stars whereas in Annie's it's in the background in places.

Olly

[Adreneline]

1 hour ago, ollypenrice said:

I might give SCNR green a spin on both of them, too. In the Pixelmath version there's a hint of green in the stars whereas in Annie's it's in the background in places.

Hi Olly,

This is probably a naive question but recognising that I map Ha:OIII:OIII to R:G:B in equal measure (as in PixelMath) is it reasonable to assume that if I sample the image in PI or PS at a significant number of places around the image then I ought to find the average for R, G and B are all equal in value.

I can understand that if I image in RGB and assign to RGB then there might be variations in each of them and the average values might differ. Now I am sure if I use mode, median or mean to represent my 'average' then I might get different results again! The beauty of Mathematics and the frustration of Statistics.

I think I understand that if I use a dslr with R:G:B in the ratio 1:2:1 then there might be an 'imbalance' in the colour. I think I understand why cmos sensors in dslr's have RGB in the ratio 1:2:1; the eye is far more sensitive to green than red or blue and so an image with RGB in the ratio 1:1:1 would not look right to our eyes. We would immediately want to add green to make it look right - I think.

Of course it may all be down to how raw data is converted to png, or jpg or tif - all of which I know nothing.

I think I've said enough to unleash a can or worms - or two - or more! Or is it all a red herring? Or am I thinking too much?

Thanks.

Adrian

 

 

[ollypenrice]

Using Photoshop's Colour Sampler tool I like to go for parity in RGB and this is what Pixinsight's DBE aims for as well, so I don't think our eye's sensitivity changes this. Some people like a slightly higher blue value, some a slightly higher red for a warmer look, but 23/23/23 in R/G/B is fine by me as measured in Ps.

Olly

[Rodd]

I think PM is better too--the biggest difference to me is the saturation--which is too high in the PM image.  But this can be easily changed.  I think the dynamic range is a bit better in the nebula in the PM (hard to tell it its due to saturation or dynamic range--call it the apparent dynamic range).

Nice image of the central bok.

Rodd

[Adreneline]

11 hours ago, Rodd said:

I think PM is better too--the biggest difference to me is the saturation--which is too high in the PM image.

Hi Rodd,

Sorry - I'm a bit confused by this - I think one should be PM and the PS but I'm not sure which is which

Adrian ?

[Rodd]

2 hours ago, Adreneline said:

Hi Rodd,

Sorry - I'm a bit confused by this - I think one should be PM and the PS but I'm not sure which is which

Adrian ?

I was saying I like the PI better (sorry), even though the saturation is a bit high (that's an easy fix.  I really botched it didn't I!?
Rodd

[Adreneline]

4 hours ago, Rodd said:

I like the PI better (sorry), even though the saturation is a bit high (that's an easy fix).

Haha. That makes sense now.  I'll get it fixed!

Thanks again.

Adrian

[Stub Mandrel]

On 12/10/2018 at 20:19, Adreneline said:

I think I understand that if I use a dslr with R:G:B in the ratio 1:2:1 then there might be an 'imbalance' in the colour. I think I understand why cmos sensors in dslr's have RGB in the ratio 1:2:1; the eye is far more sensitive to green than red or blue and so an image with RGB in the ratio 1:1:1 would not look right to our eyes. We would immediately want to add green to make it look right - I think.

The DSLR 1:2:1 ratio is a 'canard' as debayering automatically evens it up. Even with a super-pixel approach the two green pixels are added together and the R and B ones are multiplied by two.

[Adreneline]

3 minutes ago, Stub Mandrel said:

The DSLR 1:2:1 ratio is a 'canard' as debayering automatically evens it up. Even with a super-pixel approach the two green pixels are added together and the R and B ones are multiplied by two.

Thank you for the explanation.

Adrian