A dataset made up of 3D objects added to real footage with the masks of the 3D objects

I am looking for a way to test the performance of an object removal program

I am looking for a dataset (images or videos) that is made up of realistic 3D objects added to real footage. There must be a version before the object was added as well as after the object was added, so I can compare the results. Also, I would need the true mask of the object that has been added.