Seeking advice on Computer Spec for Affinity or Pixinisght

[JohnDeathridge]

Hi there new to the society and new to astrophotography to please bear with me.

Looking into pre and post image processing applications and wanted advice on computers in use and specs as looking on various websites notably  Pixinsight's own the minimum spec which is pretty demanding and expensive.

Affinity Photo as well as Deepskystacker seem less demanding but there is still processing power needed to process the images.

Heard and have be advised that Pixinsight is way to go for both pre and post image processing of deep sky images but steep learning curve and as I said earlier minimum computer specs are high.

I have also attended a great demo of Affinity Photo by Nik Szymanec which seems to have a shallower learning curve. This seems to have lower computer demands.

Could someone advise computer specs that they are using that manages to process the images in reasonable times without having to spend a fortune on a pc please as this may influence my choice on Pixinsight or not.

Many thanks all

[Carole]

I wouldn't mind a record of this too as my computers are also not up to Pixinsight, which generally I wasn't planning to use, but some of theee new "tools" are very tempting and only available on Pixinsight.

Carole

[The Thing]

I use Affinity photo 2 DSS and Pixinsight. They'll work on most pcs with enough memory (8gb on my pc and 16gb on my astro laptop). The rest of the spec depends on how much time you like to have for making tea and mowing the lawn etc.

An SSD or two speeds things along as does a CUDA capable video card (Nvidia). I've added an old one to my pc and it really speeds up some operations in PI and AP2.

[Roberto]

John

Most of the new CMOS cameras have very large arrays with tiny pixels and produce huge files (>30MB to 100MB each in size).  To image from your location without saturating stars (to avoid losing colour) you want to keep your exposures short (<=5 min for OSC, 10 min L and 20 min NB) so will end up with a lot of exposures.  I typically take >100 OSC exposures from Petts Wood to have a smooth image which is easier to process.  When using PixInsight to process so many images, you need RAM, lots of it.  So aim for >16GB at least.  A good graphics card - NVIDIA based ones - are great for some of the processing routines that depend on using neural networks such as star removal, noise removal and deconvolution.  Go for the newest graphics card with as much dedicated memory as possible (>6GB) so that those processes are done on the graphics card's memory saving you time.
My PC - which is a laptop actually - is a 32GB RAM machine with a (now old) NVIDIA RTX2060 (6GB) from ASUS.  It is 3 years old and I'm thinking about upgrading to a 64GB RAM machine.  There are headless PCs with very good specs which are much cheaper than a brand PC or laptop (I bought my laptop right at the beginning of Covid because it has two screens which I needed for office work).  Ryzen based PCs will also work with neural network based processes.

Roberto

[Roberto]

And obviously you will need a very large SSD drive if you are planning to use your PixInsight PC as your acquisition machine also (which I do).  Go for 2TB or more if you can.

Some people have a lower spec PC at their scope with enough storage for a few nights and then transfer to a high spec machine to process (over network or by removing NVMe drives for example).

Also, PixInsight is a LOT (LOT) faster under Linux.  If you do Linux, buy a headless computer with that operating system.

[Roberto]

Join the forum as there are users who post their PC specs regularly: https://pixinsight.com/forum/index.php

The best Linux distribution for PixInsight is Kubuntu Linux 22.04 LTS.   If you are IT oriented, you can run Linux under Windows (native) with a microsoft utility called WSL (Windows Subsystem for Linux):  https://learn.microsoft.com/en-us/windows/wsl/install#install 

I have run PxI successfully this way and it is faster (even with less RAM allocated as a subsystem) than Windows.

Good luck,

Roberto

[Roy]

I routinely use PixInsight (PI) to process images from a 6MP CCD camera using our desktop computer which has the following specification:
- Intel Quad Core i5 CPU;
- 32 GB RAM;
- 64 bit Win 10 OS;
- 2 x 1TB SSD drives;
- Dell 24" monitor.

This runs PI pretty well, the only times I have to wait is during Image Integration (WBPP), Deconvolution and Noise Reduction, then it's normally a case of put the kettle on and simply marvel at the sheer amount of number crunching going on. I will say thought at these times the processor is operating at 100% and all the fans are running to keep it cool so a little bit noisy. Finally the advice I received was to run PI of SSDs and not hard drives, and I think this is essential.

Hope this helps and feel free to message me if you need any further information.

Roy

[JohnH]

To image from your location without saturating stars (to avoid losing colour) you want to keep your exposures short (<=5 min for OSC, 10 min L and 20 min NB) so will end up with a lot of exposures

Hi John,

I consider myself to be a beginner still. The learning curve is very steep and I still cannot see the summit! I probably got the wrong telescope to start off with - over ambitious - but it is still enjoyable. I think.

Exposure times do depend very much on your telescope. Mine is f2.8, this means that I take a ridiculous number of subframes on my Mono camera - up to 750 for Luminance at 10 seconds each on a bright object (M42). The result can be well over 50Gb on one target.

OSC would be more manageable but the amounts of data generated can still be prodigious. This means that you need to give your computer plenty of time to run through whichever software you choose. I recently calibrated, registered and integrated 60 mono images with 60 flats and Master Darks (which I had prepared earlier) in Pixinsight using its fantastic Batch Processing script. It took about 90 minutes on my 2017 MacBook Pro (Intel Quad Core i7 at 3.1 MHz) which has 16Gb of processor memory and 4Gb Radeon graphics card.

I am trying to get to grips with Affinity Photo 2 for additional processing so that I can stop paying Adobe.

A LARGE fast SSD is a real boon. Two, one to store your captures and one for processing the data is better for me otherwise I fill up my 750Gb on board SSD in no time.

f2.8 is a fairly extreme example but it is amazing how data builds up.

Best of luck,

John

[JohnH]

I typically take >100 OSC exposures

Roberto,

How many Flat exposures do you take? Would you balance the number of Flats to the number of Lights?

I use 60 Flats and 60 Darks, should I do more?

Thanks,

John (H)

[Roberto]

I typically take >100 OSC exposures

Roberto,

How many Flat exposures do you take? Would you balance the number of Flats to the number of Lights?

I use 60 Flats and 60 Darks, should I do more?

Thanks,

John (H)

Hi John

My setup is pretty static - I don't take the camera off my telescopes for example often.  I take 10-15 frames which last me months.  In fact, my OSC flats - which are starting to degrade mostly because of sensor drift - are more than 18 months old!

Roberto

[Ivor]

A lot of good suggestions already, a key point is what do you want to use the machine for. A capture laptop doesn't need to be powerful, a processing one does or you are making a lot of cups of tea. What else will you use the processing machine for? I built a tower PC and use mine for WFH, a bit of gaming and have a couple of screens which makes processing a lot easier.
You said you didn't want to spend a fortune, this is very subjective and laptops are different to tower PCs.

To give you a scale I spent about £800 a couple of years ago and got this:
MSI B350 tomahawk motherboard
GeForce GTX 1660 Ti VENTUS 6G OC (CUDA compatible)
AMD Ryzen 5 5600X 6-Core Processor 3.70 GHz (Mid range CPU)
CORSAIR VENGEANCE RGB PRO 16GB (2x8GB) DDR4 3600
2 X Samsung 980 PRO M.2 NVMe SSD (MZ-V8P1T0BW), 1 TB (fastest SSD format)
I already had a monitor but this is a good one
OC Gaming Q27G2E - 27 INch QHD Monitor, 155Hz

The advance of this kind of setup over a laptop is it is easy to upgrade different parts, with a laptop you are more limited.

 

[Carole]

How many Flat exposures do you take? Would you balance the number of Flats to the number of Lights?

I use 60 Flats and 60 Darks, should I do more?

I do around 13 flats per filter
Darks around the same, sometimes less if I don't have that many.

Carole