Hello dear user,
Trying to reconstruct a model from such capture is not really what you should be doing, it could possible be put together but the unwritten rule for photogrammetry is to move around the object and make the camera look at it from different angles. As you’re keeping camera in the same position and just going forward it is not going to work well. Although if you would have a rig of multiple cameras covering more angles with overlaps this could possibly work out for you, even when taking a video is a bit worse than shooting images (resolution, motion blur, etc…).
PS. While importing a video you should be asked to split it in sequence of files as importing every Xth frame