Grid Repair

My panorama stitching process with PTGUI is based on either grid alignment or on the import of (papywizard) position data. The latter is based on sensor data of the actual shoot position.

Often running the optimization will tear apart the regular grid alignment. Images without control points will remain on their original grid position, whereas the other images will be shifted to an optimized position. Then you can try to level the horizon, but you will have to move the "loose tiles" by hand to their appropriate position. Coarse alignment typically is fine, however, if you have a gigapixel panorama, this can be tedious work.

After the optimization, the regular grid is torn apart between connected and unconnected (sky and ground) images

Since I am using an automated panorama head, I always work with regular image grid. And even though the grid is a little bit distorted, the regular arrangement be used to move the loose tiles (without control points) to their appropriate positions.

A little Python script for automated position alignment of unconnected images can be downloaded from my Github repository:

https://github.com/TomNaiser/Gigapixel-Tools/blob/main/LooseTileAlignment.py

Workflow:

  • Create a python project: Load the images, perform grid alignment or import papywizard data, run optimization, level the horizon

  • Save the PTGUI project file.

  • Load the PTGUI project file with the python script, and generate a modified PTGUI project file

  • Continue working in PTGUI with the modified project file

Left: Original image positions. Unconnected tiles (sky and ground) are marked red. Right: Unconnected images are aligned in well with respect to the connected tiles.

Modified PTGUI project, reloaded in PTGUI