Tanawat Wattanachinda wrote:I can see it use in tracking software as a soft constraint for the solver.
optical track + rotation from the gyro(as a soft input)
it doesn't need to be accurate, that way
The problem I mention is that rotation isn't enough information to be useful. Unless you can account for translation, you're going to be generating data that has a higher error than if you just did the optical track by itself. The gyro could simplify the creation of a coordinate system, but that's pretty trivial anyway. If you have a shoot where you know the optical track will be terrible, because of low light or massive occlusions, or a lack of rigid tracking features, you're best off using a motion capture system for the camera or using a crane that records the joint rotations.
I still think the application is for location scouting or roughly recording the placements of different cameras in a shoot. The GPS and gyro can get you enough information to orient your shoot to LIDAR or to DEM or satellite images, plus you can have 1400 clips from a documentary and be able to sort them by location or whatnot, which is great, you'll spend less time manually logging that.
This could also be great for other industries like real estate or insurance adjustment. Oh, and for ENG. Imagine you are a local news producer and a gas main blew up a building in town. You've got a crew shooting the scene, and the metadata gets compared to your archive of footage from the past few years and the software finds all the clips that could have the building in it and sorts it by similarity to the position and angle of what you just shot. Now you've found your before clip to edit into the story.