Thread Rating:
  • 0 Vote(s) - 0 Average
  • 1
  • 2
  • 3
  • 4
  • 5
Examples of SNR's
I've got a fellow who is really interested in using Skytools4, but I'm having a difficult time helping him relate the SNR values in Skytools with the SNR values he gets from PI.  Is there a gallery of an object at different SNR's that I can show him on AstroBin or similar?  Ideally it would be several types of objects (galaxy, nebulae, planetary) at SNR's from 1 to 100 as on the quality list box ...

Failing that, is there a way to relate the SNR value from Skytools4 to the PI SNR?

I'm sorry if this is a dumb question, but what do you mean by "PI?"
Clear skies,
Head Dude at Skyhound
(2022-09-26, 02:40 AM)theskyhound Wrote: Hello,

I'm sorry if this is a dumb question, but what do you mean by "PI?"

Sorry, PixInsight ... Smile
Oh, of course. I should have realized it was Pixinsight!

To answer your last question first, SNR is SNR. Its defined mathematically so there isn't any need to relate the SNR in SkyTools to that in Pixinsight. I read something once on Cloudynights where a guy was going on about how I had created my own SNR scale for SkyTools, but that's nonsense. SNR is SNR.

What I did do, was to attempt to relate the SNR numbers to common experience by putting a human experience scale on the numbers. You can see that when you select the SNR in SkyTools.


The definitions for values of SNR from 4 to 15 are commonly accepted, SNR 100 and 1000 are by definition. The ones in between are somewhat subjective, but they give you a good general idea of what they mean. Whether or not an SNR of 30 will provide acceptable results, or if you really need an SNR of 60 or more, is very much up to the individual.

The calculated SNR values come with assumptions, caveats, and things the user should be be aware of.

Unless you take an image of a flat field, there is no SNR that applies to the entire image. Normally, as signal increases, the noise stays the same, so the Signal to noise ratio is higher for the brighter parts of the image that it is for the fainter parts. So you must specify what part of the image the SNR applies to. For example, a galaxy has the highest SNR in the core, less in the spiral arms, less still in the faint outer halo.

This is why it is important to specify the part of the object you want to expose for, or in other words, the part of the object that SkyTools will calculate the SNR for:


In this example of a galaxy, the default is for the "main extent" which is the "main" part of the object, usually neither the brightest nor the faintest, but it is best to specify the part you want to consider. When you measure on Pixinsight, you want to measure the same part of the galaxy.

Obviously there are going to be brighter knots, dust lanes, etc. So this is all an exercise in approximation. The purpose isn't to get an exact SNR but to be close enough to get an exposure that is close enough.

There are also assumptions being made in SkyTools, such as the stacking is done via the most efficient algorithm. There can be poor data available for the target object (galaxies are generally very good, but HII regions may not have any emission line data at all, so the SNR result may only be approximate.). The user should study the Object Information for the object and note things like the quality of the data.

In the end, the best way to interpret SNR values between 20 and 100 is to measure the SNR on parts of your own image stacks and decide for yourself what they mean to you.
Clear skies,
Head Dude at Skyhound
[-] The following 1 user says Thank You to theskyhound for this post:
  • choward94002

Forum Jump:

Users browsing this thread: 1 Guest(s)