PixInsight- I think I'm sold!

[Icesheet]

I have been avoiding PixInsight since I started with astrophotography. Partly because of the price and partly because PS was a big enough learning curve for me along with everything else. Well, I decided to take the trial and see what I could do with existing data. M31 has been my reference point since it was my first ever attempt. In a little under 2 hours I produced, what I believe to be a better final product than all the fiddling I could do in PS. It's still not perfect, blown the core out more here, but to my eyes a definite overall improvement, especially with colour balance.

 

Let me know what you think. Looks like I'll need to dig deep for another astro related purchase! 

 

 

Edited January 29, 2021 by Icesheet
Picture not loading

[Gerr]

Love the top picture (hope this one is your improved image).

The bottom picture looks too bright (galaxy cores), noisy and too many bloated stars. Light gradient is more obvious in bottom left.

 

Gerr.

[Icesheet]

2 minutes ago, Gerr said:

Love the top picture (hope this one is your improved image).

The bottom picture looks too bright (galaxy cores), noisy and too many bloated stars. Light gradient is more obvious in bottom left.

 

Gerr.

No it’s not haha 😆 it’s the bottom one! But, I loaded this on the computer and there it looks completely different. Looking on my phone now, I agree that the top one is better.

Can I ask if you’re on a mobile device or a desktop? I’ve sometimes noticed differences before but nothing like this

[Icesheet]

Hmm I just checked on my computer.   On PI it looks good. For sure the Galaxy is a bit brighter along with background sky but overall a lot more detail and better colour. I opened the jpg I saved and it’s noticeably brighter there and here on my phone it’s even brighter again. 
 

The jpg of my earlier version on the computer lacks detail and contrast and seems to have a purple cast in the background sky. Here on the phone it actually looks much better. I have no idea what’s going on. As long as you like one of them I’m happy 😉

Edited January 29, 2021 by Icesheet

[discardedastro]

It's one hell of a learning curve - but it will let you produce some utterly fantastic results and does absolutely blow every other package out of the water in my opinion.

One common mistake with PI is leaving STFs enabled - hit F12 (you'll see the green band on the image label if you've got an STF applied). The screen transfer function effectively tries to stretch data so you can see it but can cause confusion in post-processing.

I'd heavily recommend Inside PixInsight as a book reference. It's full of good tips and clear explanations of tools - it is the missing manual.

You've definitely got more detail and background in your latter image - this is where tools like DynamicBackgroundExtraction will help a lot to draw out more contrast.

One other thing if you're aiming for consistent results on all displays - an excellent thing to get is a colour measurement sensor for your monitor and calibrate it (e.g. X-Rite i1Display). I use DisplayCAL to run the calibration and apply it. This will also give you an idea of how "accurate" your monitor is both for colours and brightness variation. This at least lets you get colour approximately "right" on your screen.

[Gerr]

I viewed them on my laptop. But now after looking at them on my mobile there is not much between them! The bottom one appears bigger and more contrasty but I still like the top one as the galaxy stands out better and the stars are less ‘fuzzy’.

Everyone has their own perception of such things but the photoshop one has it for me!

Gerr 😀

[Icesheet]

7 minutes ago, discardedastro said:

It's one hell of a learning curve - but it will let you produce some utterly fantastic results and does absolutely blow every other package out of the water in my opinion.

One common mistake with PI is leaving STFs enabled - hit F12 (you'll see the green band on the image label if you've got an STF applied). The screen transfer function effectively tries to stretch data so you can see it but can cause confusion in post-processing.

I'd heavily recommend Inside PixInsight as a book reference. It's full of good tips and clear explanations of tools - it is the missing manual.

You've definitely got more detail and background in your latter image - this is where tools like DynamicBackgroundExtraction will help a lot to draw out more contrast.

One other thing if you're aiming for consistent results on all displays - an excellent thing to get is a colour measurement sensor for your monitor and calibrate it (e.g. X-Rite i1Display). I use DisplayCAL to run the calibration and apply it. This will also give you an idea of how "accurate" your monitor is both for colours and brightness variation. This at least lets you get colour approximately "right" on your screen.

Thanks. I'll definitely look into that book. I can already see it has unbelievable functionality. I also read most people use it along with photoshop, I will need to try and find a PI/PS workflow.

 

I actually just saved as TIFF from PI and opened in PS. Wow. I nearly burned my eyes out with the core and the histogram is way to the right. I've obviously done something wrong somewhere. Lesson learned, check these things before you post them!

[discardedastro]

Just now, Icesheet said:

Thanks. I'll definitely look into that book. I can already see it has unbelievable functionality. I also read most people use it along with photoshop, I will need to try and find a PI/PS workflow.

 

I actually just saved as TIFF from PI and opened in PS. Wow. I nearly burned my eyes out with the core and the histogram is way to the right. I've obviously done something wrong somewhere. Lesson learned, check these things before you post them!

For what it's worth, I don't use PS at all for any of my images - PI can do everything I want. Probably the only thing I'd use PS for would be watermarks/signature image addition.

Depending on what your source material is you've probably stretched with HistogramTransformation while you had a STF applied. Super easy to do depending on what your starting point was (linear vs nonlinear, starting exposure levels, etc).

[Icesheet]

3 minutes ago, discardedastro said:

For what it's worth, I don't use PS at all for any of my images - PI can do everything I want. Probably the only thing I'd use PS for would be watermarks/signature image addition.

Depending on what your source material is you've probably stretched with HistogramTransformation while you had a STF applied. Super easy to do depending on what your starting point was (linear vs nonlinear, starting exposure levels, etc).

I think that is what I may well have done. Just looked back at the tutorial I was following and it mentions removing the STF autostretch before applying histogram transformation. It looks like I probably didn't. Back to the drawing board.

[Icesheet]

14 minutes ago, Gerr said:

I viewed them on my laptop. But now after looking at them on my mobile there is not much between them! The bottom one appears bigger and more contrasty but I still like the top one as the galaxy stands out better and the stars are less ‘fuzzy’.

Everyone has their own perception of such things but the photoshop one has it for me!

Gerr 😀

Thank you! I'm glad you were honest as now I know something has gone wrong somewhere and I don't need to worry about blinding everyone with my pictures 😉

[geeklee]

9 hours ago, Icesheet said:

I think that is what I may well have done. Just looked back at the tutorial I was following and it mentions removing the STF autostretch before applying histogram transformation. It looks like I probably didn't. Back to the drawing board.

Don't worry, the STF is never "applied" to your image. It's a view into what it will look like when a certain stretch is applied.  It is never actually altering your image like, say, the histogram transformation process. 

If STF is still enabled when you apply a normal stretch of any magnitude you'll typically end up with a white image in your view! At any point you just click F12 and take the STF off. Remember, the STF is always being calculated on the current image at that point in time so changes if the image changes and it's reapplied. For example, a later application of it after say a DBE or noise reduction would produce a different stretch.

The difference viewing between a monitor and modern smart phone is usually staggering on any image - pixel size and density, screen quality etc...

Linear image, STF applied:

Histogram Transformation applied, with STF still enabled:

F12 pressed:

STF applied again, although you wouldn't usually have a reason to do this when non-linear:

 

Edited January 30, 2021 by geeklee
added example images

[MartinB]

I cut my processing teeth on PS and over the past 2 years have been getting to grips with PI.  I have been through Adam Block's videos and looked at a lot of online resources such as Light Vortex.  PI is a brilliant programme but so is PS but in a different way.  It seems to be a badge of honour saying that an image was processed entirely in PI.  In fact all images present processing chanllenges, especially M31!  The linear noise reduction tools in PI are superb but I am sure some folk are over impressed PI nomenclature.    "Multiscale Linear Transform" sounds so much more powerful than  PS "Reduce noise"!  Infact PS reduce noise is a formidably powerful tool especially when modified by a star layer on top of a layer mask.  Although most of my processing is now done in PI I still find PS useful in some situations and also sometimes a great deal slicker.  If only had one processing programme it would be PI but sometimes it is nice to have more than one set of tools!

[Icesheet]

1 hour ago, MartinB said:

I cut my processing teeth on PS and over the past 2 years have been getting to grips with PI.  I have been through Adam Block's videos and looked at a lot of online resources such as Light Vortex.  PI is a brilliant programme but so is PS but in a different way.  It seems to be a badge of honour saying that an image was processed entirely in PI.  In fact all images present processing chanllenges, especially M31!  The linear noise reduction tools in PI are superb but I am sure some folk are over impressed PI nomenclature.    "Multiscale Linear Transform" sounds so much more powerful than  PS "Reduce noise"!  Infact PS reduce noise is a formidably powerful tool especially when modified by a star layer on top of a layer mask.  Although most of my processing is now done in PI I still find PS useful in some situations and also sometimes a great deal slicker.  If only had one processing programme it would be PI but sometimes it is nice to have more than one set of tools!

I’m looking forward to learning more despite my mishap! Plus I’ll always have PS anyway as my wife uses it for her photography so should have best of both worlds. Have now also ordered the Warren Keller book and the new Rogelio Andreo one so will have plenty to keep me going!

[Icesheet]

9 hours ago, geeklee said:

Don't worry, the STF is never "applied" to your image. It's a view into what it will look like when a certain stretch is applied.  It is never actually altering your image like, say, the histogram transformation process. 

If STF is still enabled when you apply a normal stretch of any magnitude you'll typically end up with a white image in your view! At any point you just click F12 and take the STF off. Remember, the STF is always being calculated on the current image at that point in time so changes if the image changes and it's reapplied. For example, a later application of it after say a DBE or noise reduction would produce a different stretch.

The difference viewing between a monitor and modern smart phone is usually staggering on any image - pixel size and density, screen quality etc...

Linear image, STF applied:

Histogram Transformation applied, with STF still enabled:

F12 pressed:

STF applied again, although you wouldn't usually have a reason to do this when non-linear:

 

I actually think I know what happened now. It seems I used the STF auto stretch on the histogram transformation process. However, for some reason this stretch pulled the midtones all the way over to the left of the data. It didn’t look like that on the image on screen though and not being familiar with the histogram on PI I didn’t notice this. At least that’s my working theory just now. 

[geeklee]

2 hours ago, MartinB said:

 "Multiscale Linear Transform" sounds so much more powerful than  PS "Reduce noise"!

I always have a chuckle at MorphologicalTransformation.  I think my process icons with this are called "StarReduction" and "AlterStarMask" or something similar!

[geeklee]

38 minutes ago, Icesheet said:

It seems I used the STF auto stretch on the histogram transformation process. However, for some reason this stretch pulled the midtones all the way over to the left of the data.

This sounds normal for a "by the book" stretch - dragging the STF onto the HistogramTransformation process and applying to your image.  The histogram will then look like this before applying (i've made the HT process really wide so you can see how far the midtones is across (it's merging on the left without zooming in):

A lot of the beginner processes out there follow this  approach.  I usually find the STF a very hard, uncompromising stretch, but it depends on your data.  

Anyway, I look forward to any future revisions! 👍

Edited January 30, 2021 by geeklee