Olfaction is an area of great interest as a new interface for digital experience. To provide rich multimodal experiences with limited amount of scent, we propose efficient olfactory ejection methods: pulse ejection and habituation ejection. These methods are based on Bayesian cross-modal model of olfaction. The pulse ejection gives off odour at regular intervals. Immediately after each ejection, the scent perception’s prior distribution changes, making it easier to perceive presence of scent. Habituation ejection gives off odour for a certain time at the beginning of a scene. As the nose gets accustomed to the scent, the confidence of the scent perception’s likelihood function becomes smaller, making it easier to create an illusion of scent by audio-visual information. An experiment in VR environment showed tendency that the methods, especially the pulse ejection, effectively reduced total ejection amount of the odour while maintaining the richness of multimodal experiences.