UKAI is moving to Facebook

Author Topic: PixInsight - Large Mosaic Method  (Read 580 times)

0 Members and 1 Guest are viewing this topic.

Offline chris.bailey

  • Administrator
  • Galactic Poster
  • *****
  • Posts: 10012
  • If at last you do succeed, don't try again
    • Rosemount Observatory
PixInsight - Large Mosaic Method
« on: 11:16:32, 02 December, 2017 »
Mosaics are a tempting, though time consuming, proposition for imaging very large objects or even groups of objects that work well together. The normal way of creating mosaics is to use image processing tools that look for matched stars on the overlap between images and then employ blunt force distortion correction to make the frames fit together based on the data pairs found in the overlap. This works fine for mosaics of a couple of frames but soon starts to struggle with frames that are being matched to other images on multiple edges. This is especially true when the base images are wide field in nature to start with i.e. those taken with a camera lens. All telescopes carry some degree of distortion but with camera lenses these tend to be compound and complex in nature due to the large number of optical elements employed. Using simple linear distortion correction on such images leads to areas of mismatch or outright failure to join panels together.

The following method has only a few steps but each are a little time consuming and some are quite computer processor and memory intensive. It employs a generated artifical star field as an un-distorted reference image to create a distortion model for the imaging system that is then used to apply second order distortion correction to the base images. The corrected images are then plate solved and the mosaic created using the co-ordinates of each image i.e. no star matching is employed.

(The screen shots are quite large but further downsampling makes the text unreadable)

Step 1 - Catalog Star Generator

A simple way of creating an undistorted reference image is to use Catalog Star Generator to provide a base image of the same dimensions, image scale and co-ordinate origin as one of the mosaic images. Choose one of the images that has a good number of stars right into the corners. For most images a Gnomic projection works well but for focal lengths below 100mm, Mercator may give better results.

Step 2 - Dynamic Alignment

The second step is a little laborious but only has to be done once for the imaging system being used.  This is the step that creates the basis for the distortion model so is worth spending some time on. The aim is to match up 100 or so well distributed stars in the reference image with the same image from the mosaic step. The first few stars will involve some manual correction but after 3 or 4 the star clicked on the reference should be automatically matched up on the actual image. Once complete drag the process icon onto the workspace to save it, the next step needs this process.

Step 3 - Manual Image Solver

Manual Image Solver is now used to create the distortion model (actually a comma separated value file saved in the same directory as the images being used). Manual Image Solver picks up the Process saved to the Workspace in Step 2. Make sure the reference image is the Catalog Star image created in Step 1. The little Samyang 135mm used here is actually not bad but close inspection of the errors image shows a number of areas where the stars are shifted quite considerably from the reference

( Step 3 a - Crop, rotate and DBE if needed)

Cropping a few pixels off the edges of each frame is a good idea to remove registration errors. If needs be rotate images to all have the same sky orientation and if there are strong gradients, a round of DBE with half a dozen well chosen samples can be worthwhile.

Step 4 - Image Solver

The next step is to plate solve each of the base frames to add World Co-Ordinate System (WCS) elements to the .fits header. Make sure the alignment algorithm is set to Polygons and the Distortion correction section includes the newly created distortion mode .csv file. Downloading the PPMX.bin catalog file from the PI website as a means of locally plate solving images is highly recommended, it's far more efficient for big images than using on-line catalogs.

Step 5 - Mosaic By Co-Ordinates

With all the images plate solved each frame can be put into its right location on an image covering the whole mosaic. The calculate button will do all the heavy lifting in terms of the data fields but manually selecting the same projection system used for distortion correction (Gnomic or Mercator) is necessary as is choosing High Quality under options. Have a cup of coffee, a beer and take the dog for a walk. For more than a couple of frames, this step can take hours rather than minutes.

Step 6 - Gradient Merge Mosaic

GMM takes each of the frames created by Mosaic By Co-Ordinates are seamlessly merges them together. Any star artefacts in the resulting image can normally be removed by increasing the Feather Radius setting

The mosaic can now be trimmed and processed as normal.

The extra effort involved in this method is probably not really worthwhile for creating a mosaic with only a couple of frames but once you get to beyond 2 x 2 it really starts to come into its own and should work for any number of tiles.
LX200|ZS70|FSQ85|FLT110|Altair DF250RC|EQ6 Pro(Rowan Belt Mod)|ParamountMX
ATIK383L+/EFW2/OAG|Lodestar|Baader 36mm LRGBHaSIIOIII
Starlight Express SXVRH16/ONAG/FW|Lodestar X2|Baader 2" Filters
Starlight Express SXVRH814/ONAG/FW|LodestarX2|Baader 2" Filters
Lunt LS60PTBF1200|DMK41|Quark Chromo
Samyang 135mm f1.8


ukbuysellRemote Imaging from AustraliaSharpSkyblank APTUKAI on Facebook
Powered by SMF 2.0.15 | SMF © 2006, Simple Machines LLC
DarkBreak by DzinerStudio. Theme modified by The UKAI Team

Page created in 0.455 seconds with 38 queries.
TinyPortal © 2005-2012