The goal of our research is to support face-to-face discussion performed by participants with differing languages. To support such a discussion, the Lumisight Table, which can provide different information simultaneously on a single screen, has been proposed. We have developed a multilingual face-to-face idea generation support system on a Lumisight Table. The advantage of the use of the multilingual function is to display the correct direction and in the participant?s native language. Each user seated around a table can see a similar image but characters in the images are rotated to each user?s direction. However, we found other problems in multilingual input and display. To solve the problem, we propose a concept of multi-layer structure on Lumisight Table.