Foto-Gorilla · Knowledge
Processing astrophotos: from stack to finished image
A clear route through processing, with careful checks rather than as many automatic steps as possible.
Background Gradients, Light Pollution and GraXpert
Background correction should not be confused with the complete removal of light pollution. GraXpert can reduce unwanted gradients and colour casts. However, it cannot recover all the image information that has been lost in noise because of a very bright sky, poor atmospheric transparency or insufficient imaging time.
A bright sky background affects image quality during capture. Additional background light produces its own photon noise and makes particularly faint objects more difficult to reveal. Even after a successful correction of the brightness gradient, the image may therefore contain more noise than a comparable exposure taken under darker skies.
Background correction plays an important role for my Seestar S50 Pro because I photograph many deep-sky objects from my observing location, where sky conditions are not ideal every night. Depending on the direction of the target, the Moon phase and the weather, individual exposures can be affected by different levels of background brightness.
During longer imaging sessions, the brightness of the sky background can change throughout the night. A target may initially be relatively low in the sky and rise higher later on. The Moon may also rise or set, while atmospheric transparency and haze change. Careful selection of individual exposures is therefore important for the quality of the final stack.
After stacking in Siril, I first examine the combined image while it is still linear. GraXpert can then be used to reduce unwanted background gradients. Ideally, this correction should take place before strong stretching, as brightness variations can often be handled more effectively in linear image data.
Colour Calibration, White Balance and Natural Star Colours
Colour plays an important role in astrophotography. It not only makes images visually appealing but can also convey information about temperature, chemical composition and physical processes in the universe. However, the colours in an astronomical image are not automatically reproduced accurately during capture. The camera sensor, filters, light pollution, white balance and image processing can all have a significant influence on the final appearance.
The aim of colour calibration is to balance the colour channels of an image so that stars and other celestial objects are represented as consistently and accurately as possible. Without suitable calibration, stars may appear too green, nebulae too violet or the entire sky background unnaturally reddish. Even small colour deviations can become much more noticeable during stretching, particularly in long-exposure deep-sky images.
An important distinction is the difference between white balance and astronomical colour calibration. Traditional white balance adjusts the colour channels using a neutral reference. In astrophotography, stars with known properties or suitable photometric reference data are often available for this purpose. Calibration aims to establish a reliable starting point for subsequent creative image processing.
Photometric colour calibration is particularly useful. It involves identifying stars within an image and comparing them with information from astronomical star catalogues. Using their known or modelled colour properties, the software can adjust the colour channels accordingly. Siril and PixInsight offer suitable methods for this. In PixInsight, Spectrophotometric Color Calibration (SPCC) is an important tool.
Before colour calibration, the sky background should be prepared as carefully as possible. Light pollution, moonlight and variations in background brightness can create colour gradients that make accurate calibration more difficult. It is therefore often useful to correct unwanted gradients first using GraXpert or suitable tools in PixInsight. Care must be taken to ensure that genuine nebula structures are not accidentally removed as part of the background.
Histogram, Stretching and GHS
After stacking, an astronomical image often appears very dark at first. Even though many minutes or even hours of exposure time have been collected, faint nebula structures, galaxies and fine details may be barely visible. This is because the image data is usually still in a linear state. The recorded differences in brightness are present, but they are not yet displayed on the screen in a way that allows our eyes to recognise all the faint structures.
The histogram shows the distribution of brightness values in an image. Dark pixels are found on the left, while bright pixels are found on the right. In a linear deep-sky stack, much of the image information is often concentrated in a narrow area near the left edge. This does not mean that no details are present there. Instead, the brightness values of the background, faint nebula structures and other dim areas of the image lie very close together.
To make this information visible, the image is processed using stretching. This involves carefully adjusting the brightness values so that faint structures become more prominent. Stretching does not create new image information; it makes the signal that has already been captured easier for our eyes to recognise.
The black point and white point are particularly important. The black point determines the brightness level below which areas of the image are displayed as completely black. If it is set too aggressively, faint nebula structures can be lost. The white point, by contrast, defines the upper limit of the brightest areas. If these values are adjusted too strongly, bright stars may become overexposed and lose their colour information. This unwanted loss of brightness information is known as clipping.
A particularly powerful technique for controlled astronomical image processing is GHS – Generalized Hyperbolic Stretch. Unlike a simple global brightness adjustment, GHS allows specific brightness ranges to be adjusted selectively. For example, faint outer regions of a nebula can be brought out while bright stars and the central areas of the object are treated more gently.
Image Sharpening, Deconvolution and BlurXTerminator
In astrophotography, image sharpness is one of the most important factors in revealing fine details. It is particularly significant for small galaxies, planetary nebulae, star clusters and lunar images, where it determines how clearly individual structures can be seen. However, perceived sharpness does not depend solely on the quality of the telescope. Atmospheric conditions, focus, tracking, sensor resolution and image processing also influence the result.
In a real optical system, the light from a star is not recorded as a mathematically perfect point. Diffraction, optical characteristics, atmospheric seeing and other factors spread its light across a small area. This distribution is described by the Point Spread Function (PSF). It is an important mathematical model of how an optical system represents point-like light sources.
A key term for assessing how stars appear in an image is FWHM (Full Width at Half Maximum). It describes the width of a star’s brightness profile at half its maximum intensity. In astrophotography, this value is often used to assess star sharpness and the quality of individual exposures. Under comparable conditions, a smaller FWHM generally indicates a more compact star image.
However, FWHM alone is not sufficient to assess overall image quality. A star may appear elongated because of tracking errors even if its measured FWHM initially seems unremarkable. Star shape, eccentricity, signal-to-noise ratio and the actual representation of fine details should therefore also be considered.
Traditional image sharpening enhances local differences in brightness to make edges and structures more prominent. However, if applied too strongly, it can produce artificial contrast, bright edges and increased image noise. Particular care is needed with faint nebula structures, as unwanted processing artefacts can easily be mistaken for genuine astronomical details.
Noise Reduction and NoiseXTerminator
Image noise is one of the biggest challenges in astrophotography. Particularly with faint nebulae, galaxies and dust structures, the desired astronomical signal is often only slightly stronger than the unwanted fluctuations in the image. After stretching, this noise frequently becomes clearly visible because the brightness differences in the background are enhanced along with the faint information from the celestial object.
An important concept is the signal-to-noise ratio, commonly abbreviated as SNR. It describes how clearly the desired signal stands out from unwanted fluctuations. The higher the signal-to-noise ratio, the more reliably faint structures can be displayed and processed. A smooth-looking image alone does not necessarily mean that an astrophotograph contains a great deal of genuine information. What matters is how clearly astronomical details can be distinguished within the recorded signal.
Several types of noise occur in astronomical images. Photon noise, also known as shot noise, results from the statistical distribution of incoming photons. Even if a light source emits light at a perfectly constant rate, its photons do not arrive in exactly equal numbers during every exposure. This natural fluctuation cannot be eliminated completely.
Another source is the camera sensor’s read noise, which occurs when the recorded image information is read out electronically. Dark current noise also plays a role. It is associated with thermally generated electrical charges within the sensor and can become more significant during longer exposures or at higher sensor temperatures.
In addition, there are fixed image patterns, hot pixels and other sensor-specific disturbances. These differ in some respects from random noise because they can appear at similar sensor positions across multiple exposures. Suitable calibration frames and careful image processing help reduce these disturbances. However, not every type of image noise can be removed in the same way.
Removing Stars and Processing Them Separately – StarXTerminator
Stars are part of almost every deep-sky image. They create a sense of depth, reveal the surroundings of a celestial object and can provide additional astronomical information through their different colours. At the same time, they present a particular challenge during image processing. While faint nebulae and galaxies often need to be brightened considerably, many stars are already relatively bright in the original stack. If both image components are processed together, stars may become excessively large, lose detail through overexposure or obscure the faint structures of the main subject.
One way to address this problem is to temporarily separate stars from background structures during processing. This approach is often referred to as a starless workflow. The aim is not to permanently remove stars from every astrophotograph. Instead, the stars and the deep-sky object can be processed independently and then carefully combined again.
StarXTerminator is a specialised tool for PixInsight that uses machine learning to distinguish stars from other image components. It attempts to create a starless image from the original exposure. Depending on the settings and workflow, a separate image containing the removed stars can also be generated.
The resulting image without stars is called a starless image. Ideally, it contains the structures of the celestial object itself, such as nebular gas, dark clouds or a galaxy, without interference from bright stars. The separated stars are often referred to as a star image. This can later be combined with the processed starless image.
It is important to understand the difference between a star image and a star mask. A star image contains the actual image information of the stars, including their brightness and colour. A star mask, by contrast, contains selection or weighting information that allows certain processing steps to be restricted to stars or kept away from them. Both techniques are useful, but they serve different purposes.