Plate Solving Calibration

The Plate Solver engine works in three steps:

  • It first cleans up the image (gradient removal and cropping)
  • It then detects star or planet shaped objects in the image.
  • It then uses these detected objects to "solve" the direction in which the camera was pointed.

Accordingly, calibration is also done in three steps.

Step 1. Clean up

Adjust the margin to crop out any foreground objects, such as trees or buildings.

Adjust the gradient removal strength to remove the effects of light pollution.

Step 2. Detector calibration

The detected objects will be marked with Magenta circles.

Aim to get about 25 to 40 objects detected, including all the prominent ones in the image.

First, adjust the threshold. Pixels fainter than this threshold will be ignored. This level should be high enough that noise and background pixels are ignored.

Second, adjust the minimum and maximum size so that the most prominent stars and planets are accepted, while unwanted objects (like trees, and image artifacts) are rejected.

Once you are satisfied with the detected objects, click Next. You will be able to come back and tweak them later.

Step 3. Solver calibration


This screen might look intimidating at first, but don't worry, the idea is simple. And you only need to do this once.

The engine will make a guess and show where it thinks the camera is pointing. A map of stars and constellations will appear super-imposed on the detected objects.

If this guess looks right, that is, the objects are very nearly matching, you can "Save" the calibration.

Otherwise, you can use touch gestures to adjust the position. You can move the guessed position, or rotate it. You can also zoom in or out by adjusting the FOV.

When the position looks nearly aligned you can use the Auto-tune button. When it says Matched the overlay will turn green.

When everything looks satisfactory, click the Save button. The calibration will be remembered until it is over-written by a new Save.