The sparse point cloud results generated by incremental SFM are incorrect

[Jiatutu]

I ran the sfm process with the following command：
openMVG_main_SfM -P -c 4 -s INCREMENTAL -i "F:\uavData\project_104\images\matches\sfm_data.json" -m F:\uavData\project_104\images\matches -o F:\uavData\project_104\images\reconstruction
The initial point cloud obtained was fine until 00000017_Resection.ply，The height of the point cloud seems to be wrong. It matches the height of the observation point

Can you tell me why this phenomenon occurs?

[pmoulon]

It happens if some resection are wrong. Can you try the global or incremental v2 pipeline on main_SfM or if your dataset has some gos data you can use some gos prior to help constraint the pose location