Question about the VIO backend and OpenVINS configuration used in GeoVINS

[Idea-in-Dream]

Hello, thank you for releasing the GeoVINS evaluation bags and map resources.

I am currently reproducing GeoVINS on the released TD07 data. I would like to clarify the visual-inertial backend used in the original GeoVINS experiments.

1. Was the GeoVINS estimator implemented based on OpenVINS, or was it an independent MSCKF-based implementation?

2. If OpenVINS was used, could you please provide the following information?

• OpenVINS version, branch, or commit hash;
• the complete estimator YAML configuration used for TD07;
• camera and IMU calibration files;
• IMU noise density and random-walk parameters;
• camera–IMU time offset and extrinsic calibration settings;
• initialization settings;
• clone/window size;
• MSCKF and SLAM feature parameters;
• FEJ, calibration, and zero-velocity-update settings;
• any source-code modifications made for GeoSLAM features and geographic updates.

3. Could you also clarify whether the released approximately 30 m DEM was the actual elevation map used for the GeoVINS experiments, or whether the paper results used a separate higher-resolution Gaofen-7 DSM?

At present, my reproduction uses OpenVINS v2.7 only as an implementation choice; I do not assume that the original GeoVINS backend was OpenVINS. The exact configuration would be very helpful for reproducing the TD07 results and distinguishing paper-defined settings from reproduction assumptions.

Thank you for your time.