A mosaic is a single large image made up of a number of smaller images. Astrophotographers use mosaics to create high-resolution images of regions that are larger than their telescope’s field of view. While planning a mosaic has its challenges, merging the resulting mosaic images into one larger image has its challenges too.

To find out how to plan a mosaic using Slooh.com, visit this link.

The rest of this article describes a technique to assemble mosaic panels into one larger image.

Result

This is a mosaic constructed out of a 2×2 mosaic of IC 1316, the Butterfly Nebula:

This image is good considering that the moon was just 65 degrees away from the nebula when I took these images. I took the images all in one night and planned the mosaic using Astro Mosaic.

Overview of Assembling a Mosaic

Assembling a mosaic isn’t easy and if you do it manually, there are a number of processes and measurements involved. Instead, this process takes the guesswork out of creating a mosaic by taking advantage of PixInsight’s plate solving capability. Once your images are plate solved, we use a built-in script to create the individual templates for each image. Once we have that, we use the GradientMergeMosaic process to automatically assemble our mosaic images into one larger image.

Using plate solving ensures your images are aligned precisely by PixInsight thereby removing the need for you to create templates and other processing artifacts that other approaches use.

The overall steps to assembling a mosaic using the technique presented here are as follows:

  • Process individual images to create final, nonlinear, images (optionally, you can keep your images linear until the very end. so you can adjust the entire mosaic when stretching, removing gradients, etc)
  • Plate solve each image using a built-in script
  • Use a built-in script to create the mosaic panels
  • Use the GradientMergeMosaic process to automatically assemble your mosaic

The rest of this article discusses the details of each step in the process.

Processing Your Images

Process each individual image that makes up your mosaic as you normally would. The idea is to produce a final, non-linear, image (see note above about keeping your images linear). You might apply processes that include DBE, ColorCalibration, Deconvolution, MultiscaleLinearTransform, HistogramTransformation, noise reduction, ColorSaturation and possibly SCNR if required.

This article provides you with a complete PixInsight workflow along with details about each step.

Since I created a 2×2 mosaic, I ended up with four final images. Save your images in XSIF format in a convenient place on your system’s hard drive.

Plate Solving Your Images

The next step in the process is to plate solve your images. Plate solving is a process that results in a solution being written to the header of your image. The solution includes details like the exact coordinates of the center of your image, its rotation, and the exact dimensions. Plate solving does not change the appearance of your image in any way. The solution written to your image’s header is used by a PixInsight script to create templates for use by the GradientMergeMosaic process.

Note: Before you plate solve, you’ll have to download some files and configure a process so that plate solving works correctly. You’ll have to download at least the first two Gaia EDR 3 files from the PixInsight software distribution site (these files are about 2.9 Gb each and the first two include a magnitude to 17.61, which is good enough for most telescopes; however, if you find that you have problems with star detection, download up to the first 10 files which cover stars to a magnitude of 20). Once you have downloaded the files, you need to configure them in the Gaia process (click the wrench icon to add your downloaded files).

You’ll need the approximate coordinates of your image along with the date the observation was taken

Open one of your images and note its coordinates that you can find in your Slooh.com FITS filename. For example, this is the name of one of my files:

201821p421436_20200808_023450_0_6myupl_l_cal_r.xisf

The RA and Dec form the first part of the filename. In this example, the RA is 20:18:21 and the Dec is +42 14 36.

Also, note the date the image was taken; the next component of the filename contains the date. In this example, the date the image was taken is 2020-08-08 (yyyy-mm-dd format).

With your image open, from the menu select Script – Image Analysis – ImageSolver.

The script acquires as much information as it can from the header of your image and in the case of a Slooh.com image, the script acquires the focal distance and Pixel size.

Ensure Active window is selected under the Target Image setting.

Next, you need to fill in the Image Parameters by supplying values for Right Ascension, Declination, and Observation date. Using the coordinates from the filename, fill in the Right Ascension, Declination, and Observation date fields. Note that if your Declination is negative, select the checkbox next to ‘s’.

Under Model Parameters, select Local XPSD server and select Gaia EDR3 (XPSD).

Click Ok to execute the plate solve.

Note the output in the Process Console. If things worked correctly you should see output similar to the following:

Image bounds:

   top-left ………….. RA: 20 20 18.003  Dec: +42 37 18.38

   top-right …………. RA: 20 16 22.412  Dec: +42 37 29.97

   bottom-left ……….. RA: 20 20 15.643  Dec: +41 53 58.19

   bottom-right ………. RA: 20 16 22.736  Dec: +41 54 09.66

If the plate solve fails, you’ll see the following in the process console:

*** Error: Unable to plate solve image: Alignment failed

This usually happens because the initial parameters are too far from the actual metadata of the image.

If you find that your plate solve fails, ensure that you are using the correct coordinates for your image. Also verify that the parameters for image scale and pixel size are correct (you can get both parameters from the FITS header; the focal length is the FOCALLEN keyword and the pixel size is the XPIXSZ or YPIXSZ keyword from the FITS header). If you find that plate solving fails at detecting stars, try to download more Gaia EDR3 files so that you cover magnitudes to around 20 (the first 10 Gaia EDR3 files).

Save each plate solved image as an XSIF file.

Create Mosaic Templates

The next step is to create four template files that we’ll use to assemble the mosaic. After this process, you’ll end up with four new files, each with one corner filled in with the corresponding image.

From the menu, select Script – Mosaic – MosaicByCoordinates.

Either add the XSIF files you saved or choose the windows of each of your images by selecting the appropriate button on the right.

You can leave all of the options in the bottom half of the window at their default values because we plate solved our images and the script acquires the information it needs from each individual file’s header.

Click Ok to execute the script; it will take a couple of minutes to complete.

At the end of the process, you’ll end up with four new files with each of them having one corner with an image in it. Each file is called your base filename with ‘_registered.xisf’ appended to it and it resides in the same folder as your original image.

Save each of the files in XSIF format since you’ll need the files for the next step.

Assemble Your Mosaic

The final step is to assemble the images from the preceding step into one larger image. For this, we will use the GradientMergeMosaic process – select it from the Process menu.

Add the four files from the preceding step.

In the bottom half of the screen are some options:

  • Type of combination: you can leave this at its default of ‘Average’
  • Shrink radius: this represents the number of pixels to reduce certain parts of your image at the seams. You may have to experiment with this value and for this image I used the value of six
  • Feather radius: This is the radius of the area that will be smoothed at the seams. Since we have overlap in our mosaic images, the transition should be smooth. You may have to experiment with this value, noting the effect it has at the seams of your image. For this image I used a value of 60
  • Black point: you can leave this at its default
  • Generate mask: you can leave this unchecked

Click the Apply Global button to execute the process. The output of this process is a single image with each of the mosaic tiles in it and in the correct position. You should not see any artifacts at the seams; if you do, try adjusting the Shrink radius and Feather radius options.

This concludes mosaic assembly. Adjust your image to taste and save it in your preferred format.

Conclusion

In this article, you learned a technique for assembling a mosaic using PixInsight’s Plate Solve and other processes. This process takes the guesswork out of assembling a mosaic because it uses the precise coordinates of each of your images to align your images.