Startools vs Pixinsight ( Startools a new way of image processing)

[Startrek (Martin)]

Found an interesting demonstration of Startools on a UK astro channel presented by the Startools developer Ivo Jager.
The first forty minutes or so is dedicated to why he developed Startools , an overview of how signal evolution tracking works and comparing Startools to the popular conventional Astrophotography processing software, Pixinsight.
The remainder of the presentation ( about an hour or so ) is dedicated to demonstrating Startools on a data set.
By the way this demo was aired about a year ago using Startools 1.7 and has recently been upgraded to 1.8 in January this year with some incredible features and improvements.
Worth viewing if your a Pixinsight user, a current Startools user or just a beginner looking for your first Astro imaging processing software
Cheers


https://m.youtube.com/watch?v=8Luu2-BMXzc

[Startrek (Martin)]

For those who are interested, attached is a PDF doc showing some translation terminology and functionality between Startools and Pixinsight

[sil (Steve)]

Yeah star tools has been around for years as have many other programs and they all work great for those able to understand and operate them to their fullest. They are all just tools, they dont magically do anything on their own and are useless in the hands of someone how has no clue what to do or willing to learn. Ive owned star tools and pixinsight for years, astro Art, nebulosity and now astropixel processor. I have the ones I rely on most for certain types of data and images but so far none are as powerful (read complex ) as Pixinsight and thats not to say Star tools or any other package is crap, but i dont think any other package can match PI in expert hands, not thatI would recommend PI to everyone, it requires some capability, patience and a ton of learning which is not present in everyone, and if you are doing actual science its just way above everything else. If all you want is pretty pictures, fine you have lots of choice and they all seem to folow their own workflow methodology which is something you may find difficult to understand and I just happened to take to pixinsight fairly easily for my needs, not to the level of many other users and i’m no expert with it and for beginners I always recommen AstroPixelProcessor for being so idiotproof and all you have to do, if anything is click on the numbered tabs in numerical order to follow the workflow plus it doesnt reword everything into a new language or require you to capture certain type of data using certain types of equipment or require tyou to use creative license in various steps.Your data has to be extraordinarily bad to get a bad output from it. The bigg letdown with all these packages I’ve found it solar system images, they at best have redimentary ways to register solar/lunar discs and closeups likewise planets and comet stacking, so it also depends what type of imaging you’re after. Star Tools has for years been powerful but I dont see anything to hype about it today different from the hype when I bought, certainly nothing to try to pull me away from the tools I use. Of course skilled users can do better using it than i can with the programs i use. But so what? Its a hobby only to me not a competition. Maybe i think the thread belong 8 years ago, not today St artools i said is old, its well suported and matured and in the right hand a great tool, but a tool is only as good as the user and thats where the impresivenes comes from, the end users.

[AdamJL]

Agree with that, Steve. There's no right tool objectively, only subjectively
I've tried them all, StarTools, Siril, Photoshop, PI, APP. They've all got strengths and weaknesses, and you just have to choose what works for you.

StarTools is great software, and I use it sometimes for specific things I find it's excellent at. I'll edit an image multiple times in different programs then layer blend the best bits in Photoshop! But that's my workflow and we've all got a different approach.

[Startrek (Martin)]

Thanks for your replies
My post wasn’t intended to persuade folk to jump across to Startools, or to put Startools up on a pedestal or it’s the be all and end all.All processing software does what it’s intended to do and that is process captured data into an image to the satisfaction of the user and others. I merely wanted to provide a window into how differently Startools works compared to popular processing software like Pixinsight.
It is a totally different beast in regard to signal evolution tracking processing.
I don’t think there’s been a previous presentation on Startools where the developer has gone through the mechanics of the software in such detail that’s east to understand. Previous presentations by Ivo and various You tubers have just been demonstrating a typical workflow which really doesn’t help a great deal.

Did you view the presentation ?
Definitely worth putting aside an hour and viewing it
Cheers

[Bassnut (Fred)]

Originally Posted by AdamJL

Agree with that, Steve. There's no right tool objectively, only subjectively
I've tried them all, StarTools, Siril, Photoshop, PI, APP. They've all got strengths and weaknesses, and you just have to choose what works for you.

StarTools is great software, and I use it sometimes for specific things I find it's excellent at. I'll edit an image multiple times in different programs then layer blend the best bits in Photoshop! But that's my workflow and we've all got a different approach.

"There's no right tool objectively, only subjectively". That's not entirely true. Many tools in astrophotography are technical in nature and not really subjective. Decon for instance is an early step on linear data in PI, in ST this process can be invoked at any time during processing. The creator of PI has stated Decon later (for various reasons) would be far too difficult to implement. ST has done this. PI is powerful to be sure, but its old school linear processes just indicate lazy software development to me.

[Startrek (Martin)]

Originally Posted by Bassnut

"There's no right tool objectively, only subjectively". That's not entirely true. Many tools in astrophotography are technical in nature and not really subjective. Decon for instance is an early step on linear data in PI, in ST this process can be invoked at any time during processing. The creator of PI has stated Decon later (for various reasons) would be far too difficult to implement. ST has done this. PI is powerful to be sure, but its old school linear processes just indicate lazy development to me.

Fred,
I guess you’ve viewed the presentation ?
You have pointed out one of the key elements of signal evolution processing in ST
Cheers
Martin

[Bassnut (Fred)]

yes i did!. Ivo is a terrible off-the-cuff speaker though, painful.
The differences between PI and ST are stunning. Despite PIs many powerful features (many ST doesn't have), to get a good final pic is just too hard in PI IMO, way too much tweaky BS. Use your PC to do the boring Shiet I say, not endless manual parameter options.

[AdamJL]

Originally Posted by Bassnut

"There's no right tool objectively, only subjectively". That's not entirely true. Many tools in astrophotography are technical in nature and not really subjective. Decon for instance is an early step on linear data in PI, in ST this process can be invoked at any time during processing. The creator of PI has stated Decon later (for various reasons) would be far too difficult to implement. ST has done this. PI is powerful to be sure, but its old school linear processes just indicate lazy software development to me.

I don't disagree there. And I love a lot about StarTool's design and ethos.
But I've just never liked the final output, hence the right/wrong tool Sometimes doing things the wrong way, brings out a subjectively better image (and yes, this is personal taste, I know)

And Martin to answer your question, yes I watched that video last year, not long after it was made.

[Startrek (Martin)]

Originally Posted by Bassnut

yes i did!. Ivo is a terrible off-the-cuff speaker though, painful.
The differences between PI and ST are stunning. Despite PIs many powerful features (many ST doesn't have), to get a good final pic is just too hard in PI IMO, way too much tweaky BS. Use your PC to do the boring Shiet I say, not endless manual parameter options.

Anyone who’s been programming since they were 8 yrs old , won a National industry accolade at 10 yrs old and openly confessed to being a nerd, would find difficulty in being an public speaker.
Maybe it was too early for him, like he said he wasn’t caffeinated ....

[Startrek (Martin)]

Originally Posted by AdamJL

I don't disagree there. And I love a lot about StarTool's design and ethos.
But I've just never liked the final output, hence the right/wrong tool Sometimes doing things the wrong way, brings out a subjectively better image (and yes, this is personal taste, I know)

And Martin to answer your question, yes I watched that video last year, not long after it was made.

Thanks Adam,
That’s what’s great about the hobby, we can choose our own pathways , so much choice
What did you think of the first part of the presentation, the mechanics of ST.
It certainly opened my eyes in regard to the nuts and bolts of signal processing
It is complicated ( mega mathematics ) but he made it look so simple and made many valid points which nobody could argue with him
One clever man !!
Cheers
Martin

[AdamJL]

Originally Posted by Startrek

Thanks Adam,
That’s what’s great about the hobby, we can choose our own pathways , so much choice
What did you think of the first part of the presentation, the mechanics of ST.
It certainly opened my eyes in regard to the nuts and bolts of signal processing
It is complicated ( mega mathematics ) but he made it look so simple and made many valid points which nobody could argue with him
One clever man !!
Cheers
Martin

To be honest the first part was music to my ears when I watched it last year. I remember a lot about how the concept of traditional sequential editing didn't really mean much in Star Tools and his reasoning made a heck of a lot of sense. The first thing I thought of was that whatever temp file he's creating to log the pixel-by-pixel changes must be huge! I never looked into that though.

It'd be great if he added calibration workflow into the software in future, but that's its own beast.

[gregbradley]

If PI had the interface of an iPhone it would be a truly awesome piece of software.

Its the interface I feel is where it is let down. Photoshop with its slider approach, does not really require much of an understanding of how the tool works as you can see the effect by moving the sliders.

Also needlessly complex terminology. You would need to spend a long time on the net getting definitions for technical words if you wanted to delve deeply into it.

Greg.

[Startrek (Martin)]

Originally Posted by gregbradley

If PI had the interface of an iPhone it would be a truly awesome piece of software.

Its the interface I feel is where it is let down. Photoshop with its slider approach, does not really require much of an understanding of how the tool works as you can see the effect by moving the sliders.

Also needlessly complex terminology. You would need to spend a long time on the net getting definitions for technical words if you wanted to delve deeply into it.

Greg.

Greg ,
Did you view the presentation, really worth looking at even if your not a PI user
The signal processing architecture of Startools really caught my eye. The user or operational interface to me was a secondary factor ( even though Startools only requires minimal key stroke operations or mouse operations to achieve a solid outcome , no levels , no curves or complex tweaks etc... )

PI has a signal processing architecture similar to the following -

Source Data> input > Operation 1 > input > Operation 2 > input Operation 3 and so on .....to final image
In PI, the input is not saved and utilised as important usable data in subsequent Operations.
In Startools all input data is saved and used between all signal processing Operations , so it keeps a permanent record of all processes , like time travel until the final noise reduction where tracking is switched off and a final processed image is completed.

Yes, this signal processing architecture did require volumes of complex and detailed algorithms to allow signal evolution tracking but the developer Ivo , did it and it works incredibly well

Cheers
Martin

[Geoff45 (Geoff)]

Originally Posted by Bassnut

"There's no right tool objectively, only subjectively". That's not entirely true. Many tools in astrophotography are technical in nature and not really subjective. Decon for instance is an early step on linear data in PI, in ST this process can be invoked at any time during processing. The creator of PI has stated Decon later (for various reasons) would be far too difficult to implement. ST has done this. PI is powerful to be sure, but its old school linear processes just indicate lazy software development to me.

If you have a stretched image you cannot do a proper deconvolution on it. You may be able to improve a stretched image by some process that resembles decon, but calling it deconvolution doesn’t make it so.
Proper decon relies on an accurate measure of the psf and stretched images lose the property of having a unique psf.

BTW there is nothing to stop you using the decon process in PI on a stretched image if you are so minded. It will almost certainly improve the image but mathematically speaking it won’t be a proper deconvolution.

[Bassnut (Fred)]

Originally Posted by Geoff45

If you have a stretched image you cannot do a proper deconvolution on it. You may be able to improve a stretched image by some process that resembles decon, but calling it deconvolution doesn’t make it so.
Proper decon relies on an accurate measure of the psf and stretched images lose the property of having a unique psf.

I don't think you know how ST works Geoff. ST tracks all the processes you have done. If you invoke decon after stretching, it applies decon to the original raw data and then reapplies all processes done after that.

[Geoff45 (Geoff)]

Originally Posted by Bassnut

I don't think you know how ST works Geoff. ST tracks all the processes you have done. If you invoke decon after stretching, it applies decon to the original raw data and then reapplies all processes done after that.

So it’s doing a proper decon. I have no problem with that.

[Bassnut (Fred)]

In fact any process can be correctly, mathematically, repeated at anytime before final noise reduction. A grunty PC is handy I must say. I don't understand why PI havent done this (apart from being "too difficult", as the PI developer has stated). Tracking processes like this also make final noise reduction very targeted, based on previous processes rather than masks etc.

[Startrek (Martin)]

Definitely worth viewing the presentation
It answers a lot of questions / comments raised so far
Ivo talks about “Kludges” and “Hacks” which is rife in traditional processing. These Kludges and Hacks try to correct , tweak , throttle and hide outputs and artifacts caused by sequence induced inefficiencies.
He’s taken processing to another level using signal evolution tracking.
It takes all the guess work and leg work away because the signal processing engine is humming away behind the scenes using complex algorithms and you always see a positive result which is mathematically correct.

[Startrek (Martin)]

Here’s a snippet from the Startools site which explains how signal evolution tracking works in comparison to traditional processing
There’s an excellent section on Deconvolution

https://www.startools.org/tracking/t...l-preservation

Worth looking at .......