Unistellar eVscope - Topaz DeNoise AI

Introduction | Sample Photos | Conclusions | Links

Archive

On this page, I compile my first experiences with the noise reduction application Topaz DeNoise AI when processing Unistellar eVscope photos.

Beforehand: Initial tests revealed that up to app version 1.2 no significant noise reduction can be observed, but from version 1.3 onwards, at least with some of the DeNoise AI models.

Note: For tests with Vespera photos siehe page Vaonis Vespera - Topaz DeNoise AI.

Introduction

On this page, I would like to present my first experiences that I collected when processing photos taken with the Vaonis Vespera with the noise reduction application Topaz DeNoise AI (hereafter DeNoise or DN). I bought this application in March 2023 and applied it at first to various photos that I had taken in the past in order to gain an understanding of the apps usefulness and "side effects".

Note: Even though Topaz recommends de-noising photos first and then process them further, I de-noised most of the images in an already processed version in order to save work (and keep the undocumented post-processing...).

DeNoise offers 4 alternative AI models for noise reduction: Standard, Clear, Low Light and Severe Noise. The Clear variant is said to be best suited for smooth surfaces, as they occur in galactic nebulae. That is why I initially used this model only. On closer inspection, however, it became clear to me that one should look each time to find out which model provides the best results (and with which settings...). Besides the effect of de-noising, however, processing with DeNoise can have further effects:

A slight reduction in the size of the stars (e.g. in the case of M 16)
A sometimes astonishing sharpening of the object (e.g. with M 42/43)
A "homogenisation" of colored/black areas (e.g. for B 33/NGC 2024)

The degree of these effects depends on the model used, and in my opinion it is worth paying attention to these effects.

By the way, you should also note that DeNoise can produce artifacts, especially the AI model Severe Noise! To check this, you should enlarge the objects sufficiently.

Note: A brief overview of DeNoise as well as examples of its "effects" can be found on page Topaz DeNoise AI (for Vaonis Vespera).

Sample Photos

eVscope

The following image examples are intended to demonstrate the effect of Topaz DeNoise AI on eVscope photos taken before app version 1.3 and those taken from app version 1.3 onwards. With app version 1.3, the image processing was seriously changed. And as you can see below, this also has an effect on the "editability" of the photos: The photos from app version 1.3 onwards clearly respond better to DeNoise (exceptions are always possible...).

Note: Click the previews for larger versions to see the differences!

App Version < 1.3		App-Version from 1.3 on

M 8 (Lagoon Nebula)		M 8 (Lagoon Nebula)

M 16 (Eagle Nebula)		M 16 (Eagle Nebula)

M 17 (Omega/Swan Nebula)		M 17 (Omega/Swan Nebula)

M 20 (Trifid Nebula)		M 20 (Trifid Nebula)

M 27 (Dumbbell Nebula)		M 27 (Dumbbell Nebula)

M 33 (Triangulum Galaxy)		M 33 (Triangulum Galaxy)

eVcope 2

Since the eVscope 2 comes with an app version from 1.3 and up, here are just a few examples of the effect of DeNoise (on the left you can see the file that is processed)!


M 16 (Eagle Nebula)		M 27 (Dumbbell Nebula)

M 42 (Orion Nebula)		M 51 (Whirlpool Galaxy)

Conclusions

With the eVscope (2), the type of internal image processing has a decisive influence on the effectiveness of the application of Topaz DeNoise AI: While photos taken with the older image processing show only minor improvements, photos taken with the newer image processing from version 1.3 on show significant improvements. The eVscope 2 was only shipped with the newer image processing, and its photos thus benefit from DeNoise, which is particularly evident in the example of M 51 (Whirlpool Galaxy).

In other words, photos of all eVscope versions should "respond" to DeNoise when they were taken with newer app versions (from V.1.3 on). However, I do not think that I will now process all the corresponding eVscope (2) photos with DeNoise again...

Unistellar eVscope - Topaz DeNoise AI

Introduction

Sample Photos

eVscope

App Version < 1.3

App-Version from 1.3 on

eVcope 2

Conclusions

Links