Digital fingerprinting is a technique to protect digital contents from illegal reproduction and redistribution by marking unique information for individual user. A powerful but simple attack to diminish fingerprint signals is averaging. While several fingerprinting schemes against collusion attacks were proposed, they often do not fully account for multimedia data. In this paper, we propose a zero-based code modulation method which fully reflects marking assumption concept to embedding and detection of signals and anti-collusion code working mechanism as well. By manipulating 0-bit information of binary codes, detection accuracy and code separation under averaging attacks were enhanced. To demonstrate our method, we used an averaging-resilient fingerprint code based on GD-PBIBD theory and applied it video fingerprinting systems. Through the experimental results, we convince our method improves averaging resiliency of fingerprinting systems.