Photos not calibrated

Good morning everyone,

so im working on 3 projects right now and pulling my hair out as I go. One of my jobs its showing that 450 out of 851 photos are not calibrated. Im using DJI P1 camera to do this job. This is the Smaller of the 3 jobs i tried using user tie points. but even with those photos its not helping.

  

851 photo(s) in 2 photogroup(s), 38.1 gigapixels
401 photo(s) in the main component
401 known position(s) and 401 known rotation(s)
12 control point(s) (12 full point(s), 0 horizontal point(s), 0 vertical point(s)) among which 0 check point(s)
9 user tie point(s)
257655 automatic tie point(s) View
Resolution ranges from 0.013 meters to 0.013 meters
Block has blockwise color equalization
Splats: available. Recreate

A. Positioning/georeferencing:
- Adjustment constraints: control points and photo positioning metadata
- Final rigid registration: control points

B. Main settings:
- Key point density: Normal
- Tags or QR Codes extraction: Disabled
- Pair selection mode: Default
- Component construction mode: OnePass
- Blockwise color equalization: Enabled
- Splats: Enabled

C. Estimation policies:
- Tie points: Compute
- Position: Compute
- Rotation: Compute
- Photogroup estimation mode: OnePass
- Focal length: Adjust
- Principal point: Adjust
- Radial distortion: Adjust
- Tangential distortion: Adjust
- Aspect ratio: Keep
- Skew: Keep
- Estimation groups: PerPhotogroup

Parents Reply Children
  • Hi Felix,

    Could you explain what exactly action CC performs when we use each of U17 presets?
    Alternate engine

    Complete incremental engine

    Do not check rigid registration reliability

    Do not use pose metadata for SfM

    Keep all components for adjustment

    Simple incremental engine

    Thank you!

  • Hi Elena,

    A definition of ContextCaptures AT Definition Settings Presets:

    Alternate engine: This was a brand-new engine introduced in Update16, which is more robust but slower and less scalable

    Simple incremental engine: all calibrated photos and tiepoints will remain unchanged but some orphan photos might be added to the main component.

    Complete incremental engine: This is a kind of “normal” AT but instead of computing poses from scratch, it starts from the current solution. It might add some orphan photos, merge some disconnected components and everything will be adjusted, including previously calibrated photos and tiepoints.

    Do not check rigid registration reliability: On very small acquisitions, smaller than the expected GPS accuracy, the engine may not georeference the scene. This option should force it to georeference the result in almost all cases with position metadata

    Do not use pose metadata for SfM: when computing the poses (position/rotation policies to compute) the engine may leverage the pose metadata if any to check for proper computation. If, for some reason, the pose metadata (more likely the rotations) are wrong (not properly imported for example) it can break the AT. In such cases, you can either clear the rotation metadata or use this preset to solve the issue.

    Keep all components for adjustment: If you set the poses policy to “keep” or “adjust within tolerance” this preset allows you to keep all the photos, even if they actually are several components that are not connected through tiepoints. A typical case is two sides of a river.