In this paper, we propose a pointing method for remote Mixed Reality (MR) cooperative work which an expert can instruct with captured still worker's view video. Much research has been devoted to the study of remote cooperative work, and recently systems using MR have been investigated. In these systems, users can work while looking at instructions which are superimposed as virtual objects by MR. Worker's view video is often used because an expert needs to know the situation of the workspace in real-time and exactly. But in this case, the expert's view depends on the worker's viewpoint, and the indicating point shifts if the worker moves his head or work objects. Also if an expert sees workspace through Head Mounted Display (HMD), he tends to feel sick by shaking video which is brought by the worker's head movement. In our system, an expert can stop worker's view video and direct on its captured still image. So an expert can instruct without being influenced by the moving video. It is expected to improve pointing accuracy and expert's sickness.