← Older edit
→‎Processing: stacking doc
Marked this version for translation
 
(55 intermediate revisions by 2 users not shown)
Line 1: Line 1:
=Manual page for Siril=
<languages />
<translate>


[[Siril]] is an astronomical image processing tool, see [[Siril|main page]] for more details.
<!--T:44-->
<span style="color: #ff0090; font-weight: bold; text-align: center; margin: 2em; font-size: 125%;">This page is the old documentation for early versions of Siril 0.9. Please refer to the new documentation on [https://siril.readthedocs.io/en/stable/ siril.readthedocs.io] or the new tutorials on [https://siril.org/tutorials siril.org] instead.</span>


==Convert your images in the FITS format Siril uses (image import)==
=Manual page for Siril= <!--T:1-->


To process your images with Siril, you must convert them to the FITS format it uses (16-bit unsigned, bottom-top order, 1 or 3 axes). Fortunately, Siril is able to convert some image formats to this format, mainly RAW BMP and NetPBM binaries for images, AVI for films.
<!--T:2-->
This was the main documentation page for [[Siril]], a free software for astronomy image processing. It contains several video tutorials and links to other pages that contains pictures or videos. Many features are documented by a video. The processing workflows for deep-sky and planetary are documented as a series of pictures with text or as a film in the [[#Tutorial_for_a_complete_sequence_processing|complete sequence processing]] section.


First, the working directory has to be set. Images will be taken from and converted to this directory. Click on "Change dir" at bottom right of the window to change it.
==A powerful astronomy image viewer== <!--T:3-->


The fill the boxes for source and destination file names for the conversion. Here is a screen capture of the conversion tab, where a sequence of images called <code>IMG_{number}.CR2</code> is being converted to <code>m27_{same_number}.fit</code>.
<!--T:4-->
A Display Mode viewer is used to improve the visibility of an image, but without altering its pixel data in any way, similarly to what [[DS9]] scale does.
*Linear: the default mode in Siril. Pixels are displayed from the darkest to the brightest in a linear scale.
*Logarithm: the logarithmic scale. The operation emphases simultaneously information in faint and bright levels of the image.
*Square root: the square root of each pixel. The result is close to the logarithm mode's.
*Squared: squared of each pixel. What can be seen with this mode is mainly the brightest part of the image.
*Asinh: the inverse hyperbolic sine is commonly used, it reproduces the perceptual ability of the human eye, which can accommodate to perceive dramatically different levels of brightness simultaneously. Asinh is close to the logarithm mode but has a better behaviour around zero.
*AutoStretch: Siril performs an automatic screen stretch.
*Histogram: the histogram equalisation. It increases the contrast of the image by increasing the dynamic range of intensity given to pixels with the most probable intensity values. It is very valuable to evaluate all the signal contained in the image.


[[File:Siril_conversion_screen.png]]
<!--T:5-->
All these modes can also be applied independently on each layer, when layer chaining is disabled.


As you can see, the 1 axis/plane/layer format has been deactivated because it has not been tested since the rewriting of Siril, so conversion only works with 3-layer RGB files for now. Conversion of RAW files uses dcraw without colour or gamma correction, and that can take quite some time.
<!--T:6-->
This video tutorial shows the different modes you can use to display your images.


==Work on a sequence of converted images==
<!--T:7-->
<html><video controls width="750">
    <source src="/videos/Display.webm" type="video/webm">
    Sorry, your browser doesn't support embedded videos.
</video></html>


Once you have your Siril's format .fit files, you can switch to the ''sequence'' tab seen below. Sequences are what Siril uses to manipulate files.
<!--T:8-->
In addition to this, Siril provides different colour maps display, including false colour rendering.
[[File:Siril_colormap.png|600px]]


[[File:Siril_sequence_screen_empty.png]]
==Tutorial for a complete sequence processing== <!--T:29-->


Click on the "search sequences" button. If you only have one .fit files sequence in the working directory, it is automatically selected. Else, the list will pop-up an you'll be able to select the sequence you want to use. When the sequence is loaded, two windows are opened: the grey frame control window, on the left of the image below, and the RGB composition rendering window, behind the main window on the image below.
<!--T:30-->
An illustrated tutorial has been compiled to describe the usual processing steps, starting from RAW DSLR images to stacking.


[[File:Siril all windows seq load.png]]
<!--T:31-->
* [[Siril:Tutorial_import|Convert your images in the FITS format Siril uses (image import)]]
* [[Siril:Tutorial_sequence|Work on a sequence of converted images]]
* [[Siril:Tutorial_preprocessing|Pre-processing images]]
* [[Siril:Tutorial_manual_registration|Registration (Global star alignment)]]
* [[Siril:Tutorial_stacking|Stacking]]


The grey window is used to display colour components for the image, control their composition in the colour image, and give information about the displayed frame, like the layer's name (for now it can be only Red Green or Blue, but it will be settable to narrow bands some day), pixel value when moving the mouse on a pixel, file name and layer number, FWHM calculation is an area is selected (result is quite experimental), and zoom value.
<!--T:32-->
A video tutorial is also available to describe the usual planetary processing.


[[File:Siril grey window.png]]
<!--T:33-->
<html><video controls width="750">
    <source src="/videos/mars_processing.webm" type="video/webm">
    Sorry, your browser doesn't support embedded videos.
</video></html>


The cursors at the bottom control thresholds for black and white on the grey image, allowing to play with contrast and lightness of each channel, mainly for previewing purposes.
==Process your single images== <!--T:9-->


==Pre-processing images==
<!--T:40-->
It is one thing to process a sequence of images, and another to work on only one image, generally the result of this processing. This section documents the most useful features for one-image processing.


One a sequence is loaded, images can be pre-processed, registered and stacked. The pre-processing is an optional step and involves offset, dark and flat frames. If you don't know what this is, you don't need it, but I suggest you read about it somewhere. Only one frame can be given for each of these three, and it has to be in Siril's FITS format. They should be created by median stacking, so you need to convert the source images to FITS, load the sequence and stack them before using them here. We'll assume this is done. For more info about stacking, see the [[Siril:Manual#Stacking]] section below.
===Background extraction=== <!--T:41-->


The pre-processing does not overwrites original FITS files from the sequence but creates new FITS files with the pre-processed data. The new file names are the same than the original prefixed with pp_ if you leave the default value in the prefix box. Below is an example of pre-processing using offset and dark files.
<!--T:10-->
When an image contains an intense light pollution, it is useful to remove it. Siril provides a "Background Extraction" tool in order to remove the gradient of the sky background. Siril uses a polynomial up to the fourth degree. The 4th degree polynomial is generally a good choice, for this model, though the tool allows for first to fourth degree.


[[File:Siril preprocessing screen.png]]
<!--T:43-->
<html><video controls width="750">
    <source src="/videos/BackgroundExtraction.webm" type="video/webm">
    Sorry, your browser doesn't support embedded videos.
</video></html>


==Registration (manual image alignment)==
<!--T:12-->
After extraction you can switch between image and background view in order to check the result. If you are satisfied you can thus apply the correction (subtraction or division). The subtraction operation is very different from the simple operation used for the darks and bias/offsets. Indeed, as the level of the background is identical to the one of the image, applying a simple subtraction would result in a image with a lot of black pixels. To avoid any inconvenience, the operation is done with 32-bit signed integers and finally converted into native format (16-bits unsigned).


Siril has an automatic registration capability, but I can't say it's working fine at all. I suggest you use it as a first step if the quality of your images is good, or not at all if stars are not round, if focus is bad or if there is a lot of noise. Currently, only translation is used for image alignment, no rotation.
<!--T:13-->
Following, respectively, the start image and final image (after background extraction with subtraction and equalization):


The first step is to load the sequence, if we continue this tutorial this will be the pp_m27 sequence. Go back to the sequence tab, and click on "search sequence" to find the new sequence created by the pre-processing. Select the pp_ sequence, which will open the image preview windows if not already opened, and display the first image of the sequence.
<!--T:14-->
[[File:BKG_before.png|600px]]


[[File:Siril load pp sequence.png]]
<!--T:15-->
[[File:BKG_after.png|600px]]


[[File:Siril pp seq loaded.png]]
===Color calibration=== <!--T:16-->


From this tab, you can change the image being viewed in the grey and colour windows, include or exclude them from a user-defined selection that will later be used to indicate which files should be stacked or not, and set the reference image. While browsing through the images of the sequence, a reference image may be set to a specific image if you think it has a better quality than the others, or if it's best centred. Note that when the reference image is set, it will be saved in the sequence and it will be opened first when loading the sequence the following time.
<!--T:17-->
The colour calibration tool performs a balancing of the colours of a linear RGB deep-sky image. This video tutorial shows how to use the tool. Note that Background Equalization (now called Background Neutralization) has been moved in the calibration box because it is needed before any calibration.


An alternative to the use of buttons of the sequence tab is to use the ''sequence list''. Double click on the grey window to make it appear:
<!--T:18-->
<html><video controls width="750">
    <source src="/videos/Color_Calibration.webm" type="video/webm">
    Sorry, your browser doesn't support embedded videos.
</video></html>


[[File:Siril sequence list.png]]
===Reducing green tint in processed images=== <!--T:19-->


You can then open an image of the sequence by double-clicking on its line, and select (include or exclude) the image from the sequence by checking or unchecking its box. X Y and FWHM columns will be displayed later, when registration data is available. In the image above, you can see that two images have already been excluded from the sequence.
<!--T:20-->
This video tutorial shows how to remove the green tint in processed images using the graphical user interface. This function can also be accessed from the command line, with the [[Siril:Commands#rmgreen|<tt>rmgreen</tt> command]].


To do the automatic registration, select a star in a grey image, preferably in the same layer than the layer used for registration (green by default) to be sure that the star appears in this layer, and click the "Go register" button.
<!--T:21-->
<html><video controls width="750">
    <source src="/videos/RemoveGreen.webm" type="video/webm">
    Sorry, your browser doesn't support embedded videos.
</video></html>


[[File:Siril auto registration screen.png]]
<!--T:22-->
===Fast Fourier Transform (FFT)===
A FFT algorithm is integrated to Siril in order to provide you a tool to evaluate fixed pattern noise of an image. You can thus remove the frequency pattern of the noise in the modulus file and save the result. Modulus and phase are saved in different files so you can compute the inverse transformation to retrieve the corrected image.


[[File:Siril registration done area selected.png]]
<!--T:23-->
<html><video controls width="750">
    <source src="/videos/FFT.webm" type="video/webm">
    Sorry, your browser doesn't support embedded videos.
</video></html>


You can view the registration data in the list on the left when you are viewing the layer corresponding to the registration layer. You can also see in the image above that the reference image is on the yellow line and the currently displayed image has its name in bold.


----
<!--T:39-->
===Tutorial for a complete image processing===
An illustrated tutorial has been compiled to describe the usual [[Siril:Processing_tutorial|processing steps]] for a single image processing, starting from the stacked file.


Now, in order to validate the registration, which is never perfect, there is a way to display the reference image over the other images, taking into account the translation data, and also adjust this data to have a better fit. This is done in the preview area of the main window, in the registration tab.
<!--T:24-->
==RGB Compositing==
A new channel synthesis tool is provided in Siril. This tool allows you to mix and align up to 7 channels plus one for Luminance. The tool handles the binning, just make sure to not crop the picture before: the bin1x1 MUST be loaded first to give the composited image dimensions.


Click on "Set first preview", then click on a star in the image when you have loaded the registration layer of the reference image. The preview area of the image in native resolution will be displayed in the main window. Do the same for the second preview, on the opposite of the image if several stars are available.
<!--T:25-->
The first video shows a mixing between L, R, G and B channels with same size and no alignments needed: the L channel is given by an H-alpha processed image.
This second video shows mixing between images with different binning. The alignment is also required.  


[[File:Siril manual registration regions.png]]
<!--T:26-->
After mixing, save your result and process it as you want.


[[File:Siril_registration_previews.png]]
<!--T:27-->
<html><video controls width="750">
    <source src="/videos/RGB_compositing.webm" type="video/webm">
    Sorry, your browser doesn't support embedded videos.
</video></html>


Now when you select an image in the sequence, you will see this image's layer's preview appearing with optionally the reference image's preview in transparency. Verify that both stars are aligned, and if not, use the X translation and Y translation spins to change it (use the mouse wheel or the arrows, it's easier). When the automatic registration doesn't work well, you have that kind of result at first:
<!--T:28-->
<html><video controls width="750">
    <source src="/videos/RGB_compositing2.webm" type="video/webm">
    Sorry, your browser doesn't support embedded videos.
</video></html>


[[File:Siril regpreview not aligned.png]]
==Astrometry== <!--T:34-->


If you can only have one star well aligned like the first one below and the other star way too far from the reference star like the second one below, it means that the camera has turned between the shots, and currently Siril is not able to rotate the frames, so you can exclude it from the sequence.
<!--T:35-->
Siril provides a few astrometry tools.


[[File:Siril regpreview aligned1.png]]
<!--T:36-->
In particular, the PSF function that uses the [https://en.wikipedia.org/wiki/Levenberg%E2%80%93Marquardt_algorithm Levenberg–Marquardt algorithm] for minimization.


You should be able to get that kind of result if camera was not rotated:
<!--T:42-->
See the [[Siril:PSF|dedicated page]] for a video tutorial.


[[File:Siril regpreview aligned2.png]]
</translate>
 
One you have done it with all images, you will have the final list of selected images from the sequence, and much better alignment data than with the automatic registration:
 
[[File:Siril_registration_done.png]]
 
==Stacking==
 
The final thing to do with Siril is to stack the images. Go to the "processing" tab, indicate if you want to stack all images or only selected images (the third option is not yet implemented). Then simply click on "Start summing stack" which is the regular stack. The median stack is mostly used to create pre-processing frames, like offsets or darks.
 
[[File:Siril stacking screen.png]]
 
After that, the result is saved in the file named below the buttons, and is displayed in the grey and colour windows. You can adjust levels if you want to see it better, but that will not be saved in the FITS file.
 
[[File:Siril stacking result.png]]
 
An export of the RGB composed image is being implemented, in order to use Siril's composition and level adjustment, but I recommend using an external tool such as the wonderful [[DS9]] to adjust your levels. In particular, I use the asinh scaling for blue and green data because they are generally much less equalized than the red when using an unfiltered DSLR camera. Then, you can export the image and adjust final contrast and lightness in The Gimp for example.

Latest revision as of 22:24, 16 September 2023

Other languages:

This page is the old documentation for early versions of Siril 0.9. Please refer to the new documentation on siril.readthedocs.io or the new tutorials on siril.org instead.

Manual page for Siril

This was the main documentation page for Siril, a free software for astronomy image processing. It contains several video tutorials and links to other pages that contains pictures or videos. Many features are documented by a video. The processing workflows for deep-sky and planetary are documented as a series of pictures with text or as a film in the complete sequence processing section.

A powerful astronomy image viewer

A Display Mode viewer is used to improve the visibility of an image, but without altering its pixel data in any way, similarly to what DS9 scale does.

  • Linear: the default mode in Siril. Pixels are displayed from the darkest to the brightest in a linear scale.
  • Logarithm: the logarithmic scale. The operation emphases simultaneously information in faint and bright levels of the image.
  • Square root: the square root of each pixel. The result is close to the logarithm mode's.
  • Squared: squared of each pixel. What can be seen with this mode is mainly the brightest part of the image.
  • Asinh: the inverse hyperbolic sine is commonly used, it reproduces the perceptual ability of the human eye, which can accommodate to perceive dramatically different levels of brightness simultaneously. Asinh is close to the logarithm mode but has a better behaviour around zero.
  • AutoStretch: Siril performs an automatic screen stretch.
  • Histogram: the histogram equalisation. It increases the contrast of the image by increasing the dynamic range of intensity given to pixels with the most probable intensity values. It is very valuable to evaluate all the signal contained in the image.

All these modes can also be applied independently on each layer, when layer chaining is disabled.

This video tutorial shows the different modes you can use to display your images.

In addition to this, Siril provides different colour maps display, including false colour rendering.

Tutorial for a complete sequence processing

An illustrated tutorial has been compiled to describe the usual processing steps, starting from RAW DSLR images to stacking.

A video tutorial is also available to describe the usual planetary processing.

Process your single images

It is one thing to process a sequence of images, and another to work on only one image, generally the result of this processing. This section documents the most useful features for one-image processing.

Background extraction

When an image contains an intense light pollution, it is useful to remove it. Siril provides a "Background Extraction" tool in order to remove the gradient of the sky background. Siril uses a polynomial up to the fourth degree. The 4th degree polynomial is generally a good choice, for this model, though the tool allows for first to fourth degree.

After extraction you can switch between image and background view in order to check the result. If you are satisfied you can thus apply the correction (subtraction or division). The subtraction operation is very different from the simple operation used for the darks and bias/offsets. Indeed, as the level of the background is identical to the one of the image, applying a simple subtraction would result in a image with a lot of black pixels. To avoid any inconvenience, the operation is done with 32-bit signed integers and finally converted into native format (16-bits unsigned).

Following, respectively, the start image and final image (after background extraction with subtraction and equalization):

Color calibration

The colour calibration tool performs a balancing of the colours of a linear RGB deep-sky image. This video tutorial shows how to use the tool. Note that Background Equalization (now called Background Neutralization) has been moved in the calibration box because it is needed before any calibration.

Reducing green tint in processed images

This video tutorial shows how to remove the green tint in processed images using the graphical user interface. This function can also be accessed from the command line, with the rmgreen command.

Fast Fourier Transform (FFT)

A FFT algorithm is integrated to Siril in order to provide you a tool to evaluate fixed pattern noise of an image. You can thus remove the frequency pattern of the noise in the modulus file and save the result. Modulus and phase are saved in different files so you can compute the inverse transformation to retrieve the corrected image.


Tutorial for a complete image processing

An illustrated tutorial has been compiled to describe the usual processing steps for a single image processing, starting from the stacked file.

RGB Compositing

A new channel synthesis tool is provided in Siril. This tool allows you to mix and align up to 7 channels plus one for Luminance. The tool handles the binning, just make sure to not crop the picture before: the bin1x1 MUST be loaded first to give the composited image dimensions.

The first video shows a mixing between L, R, G and B channels with same size and no alignments needed: the L channel is given by an H-alpha processed image. This second video shows mixing between images with different binning. The alignment is also required.

After mixing, save your result and process it as you want.

Astrometry

Siril provides a few astrometry tools.

In particular, the PSF function that uses the Levenberg–Marquardt algorithm for minimization.

See the dedicated page for a video tutorial.

Retrieved from "https://free-astro.org/index.php?title=Siril:Manual&oldid=8499"