We’re encountering bowl-shaped distortion in DSMs generated from Parrot Anafi AI flights over our mill yards. We’ve tested both Agisoft and Pix4D Capture workflows and don’t see this issue with other drone brands. We’ve tried adjusting parameters like f , cx , and cy , which sometimes helps mitigate the distortion—but it’s mostly guesswork, and the same drone with a different flight mission can produce very different results. Has this issue been reported by other users, and are there recommended steps or best practices to consistently avoid this distortion?
Hello @JBojko,
To better understand your case, could you share which drone version and software release you are working with? That would help us check whether this matches a previously identified scenario.
This type of issue was known in the past and led to a detailed analysis, followed by the implementation of an optical center calibration process on Anafi Ai.
If you are still seeing bowl-shaped distortion today, it is most likely related to an imperfect calibration of the optical centers (unless you are using one of the very first hardware batches or an early software version).
As best practices to mitigate the effect:
- Plan flights with double grid patterns instead of single grids.
- Avoid manually optimizing camera intrinsics (focal length, optical center, etc.) and rely on the values provided in the image metadata.
Best regards,
Hugo
Hello Hugo,
thank you a lot for you response. We always instruct our pilots to use the newest version of the software, and we also bought a new Parrot Anafi last month. So I believe we have a very new model. Still, we are experiencing the issue regardless of the software or drone version. Are you saying that the newest updates fixed the issue ?
Best regards,
Jan
Hello,
Just to clarify, are you seeing this issue on all your Anafi Ai units, or only on a single drone?
Normally, this problem should not occur. It was identified early in development and led to a specific optical center calibration process. All production units should have gone through this calibration, so if you are using a recent and up-to-date drone, this could indicate a calibration defect on the unit itself, in which case a replacement might be needed.
As a workaround to reduce the effect, you can:
- Plan flights with a double grid instead of a single grid.
- When optimizing camera parameters, allow optimization of tangential distortions but avoid optimizing the optical centers.
Best regards,
Hugo