Guidelines for data added to AIM models

Follow these guidelines when you check the data before adding it to or creating an AIM model.

• Make sure that the subject in the file is moving. The movement does not have to be exactly the same as the rest of the captures, just that it includes most of the movement. When you add more measurements to an AIM model, it becomes less important that the added files include a lot of movement. You can verify that the movement is large enough by looking at the AIM bones, see section How to verify and edit AIM bones.

  Even if you want to make an AIM model for a static measurement it is important to have some movement in the first file. This is to make sure that the internal bone definitions are correct. However, the subsequent files can be static because then the definition is already correct.

• Make sure that the trajectories have the correct identity throughout the file. You can select a smaller measurement range to delete the incorrect parts if you do not want to identify the whole file.

• The colors of the trajectories and any bones between them are also saved in the model. For example to make it easier to verify the identification after AIM has been applied, the colors of the trajectories can be set with Set different colors on the Trajectory info window menu before creating the AIM model.

• Trajectories that are left unidentified or discarded will not be included in the model.

• If you have several subjects in the same measurement it is recommended to make an AIM model for each subject, see chapter AIM models for multiple subjects in the same measurement.

The following two steps are not as important when you add files to the AIM model.

• The AIM model will be better if all of the trajectories are visible throughout the whole measurement. Therefore it is a good idea to gap-fill the trajectories as long as the gaps are relatively small. However if there are large gaps it is sometimes better to omit the gaps by limiting the measurement range in step 2 below. This is especially important if you use clusters, because then you have several trajectories that move along the same path.

• The data of each trajectory should be as good as possible. This is especially important if your model includes trajectories that are close to each other. Then a small erratic noise on a trajectory can make two trajectory paths come very close to each other, which makes the identification difficult. Therefore if the data is erratic, i.e. the trajectory movement is not smooth when you play the file, you should delete the erratic data by the following process:

  • Find the frame where the erratic data starts. Split the trajectory at that frame, see chapter Split part after current frame.

  • Then step through the frames to locate where the trajectory data is OK again. Split the trajectory again.

  • Delete the part with erratic data that you have just created.

  • Repeat these steps for all of the frames where you can find erratic data and then gap-fill the trajectories