Hello RealityCapture Community,
I am working on a high-precision documentation project of a 20-meter historic wall. My goal is to produce a 2D orthophoto (elevation) that can be measured in CAD with sub-2mm accuracy.
My challenge is the final alignment of the 3D model. A simple visual alignment (like using the “Define Ground Plane” tool or manually rotating the component) is not sufficient, as it introduces rotational errors that are far outside my 2mm tolerance.
I need a robust, instrument-based workflow. I am already using many coded targets to ensure the model’s internal scale and geometry are accurate, but I need help with the final orientation.
My specific questions are:
What is the recommended professional workflow to define a local coordinate system (e.g., a precise XZ-plane for the facade) to achieve survey-grade (<2mm) orientation?
I assume I must use GCPs measured with a Total Station. If I measure 3 (or more) of my coded targets (e.g., Origin, X-axis, Z-axis) with a Total Station, what is the exact procedure to import these coordinates and set the “local:1 - Euclidean” system in RC to force this high-precision orientation?
How does RC handle this? Does it perform a 7-parameter (Helmert) transformation to “fit” the entire component onto these few high-accuracy GCPs?
Will this create a conflict with the other coded targets that are not measured with the Total Station and are only used for scaling? What is the best practice for this “hybrid” setup?
Is this approach overkill? I am wondering if I am overcomplicating this with the Total Station requirement. Is there a simpler, more straightforward (and perhaps less expensive) method to reliably achieve this <2mm alignment accuracy that I might be missing?
Thank you for sharing your expertise. 