Support Website Contact Support Blog

Step 1 Re-Optimization?


Quick question on Step 1 Reoptimization after marking GCPs and Check Points. After this step, can I switch all the GCPs I had marked previously as GCPs to “Check Points” to see how close my PPK data is on it’s own (with no GCPs) or would the check point results in the Quality Report be skewed because I already did Step 1 and Reoptimized after adding initial GCPs/Check Points? If this is the case, I assume I would have to Rematch and optimize to get a more balanced comparison against check points?


Here’s how I assess my PPK data.

  1. Load images geotagged with PPK data
  2. Import GCPs but don’t mark them on photos
  3. Process step one
  4. Mark GCPs on images using raycloud and set them as check points. As you mark them, you’ll be able to see the accuracy of each point.
  5. Reoptimize and regenerate quality report.

By using this method, the GCPs will never be used in the model so you can get a true accuracy assessment of your PPK data. If the result doesn’t meet your accuracy needs, then convert one point at a time from checkpoint to GCP and reoptimize. Then you can see how many GCPs you actually need to meet your accuracy requirement with PPK.

1 Like

Thanks Andrew.

Based on my workflow above…marking GCPs and Check Points and Re-optimizing after I run Step 1. It doesn’t sound like I can toggle the GCPs I initially marked before to check points status and re-optimize again without getting results that would be skewed. Meaning because I marked those GCPs initially and classified them as GCPs, they are still somehow held as GCPs even though they’re now considered check points?



Andrew’s method gives you the possibility to check the actual accuracy of your PPK data, which weren’t affected by any other external source of information - GCPs.

Looking at the workflow you applied, marking the GCPs changed initially implemented PPK information.

By changing the label of GCPs to checkpoints you won’t remove that impact the GCPs had already had on your dataset.

Therefore, I would recommend creating a project from scratch and following the workflow Andrew mentioned.