Abstract
Recent advancements in human motion synthesis have focused on specific typesof motions, such as human-scene interaction, locomotion or human-humaninteraction, however, there is a lack of a unified system capable of generatinga diverse combination of motion types. In response, we introduceSitcom-Crafter, a comprehensive and extendable system for human motiongeneration in 3D space, which can be guided by extensive plot contexts toenhance workflow efficiency for anime and game designers. The system iscomprised of eight modules, three of which are dedicated to motion generation,while the remaining five are augmentation modules that ensure consistent fusionof motion sequences and system functionality. Central to the generation modulesis our novel 3D scene-aware human-human interaction module, which addressescollision issues by synthesizing implicit 3D Signed Distance Function (SDF)points around motion spaces, thereby minimizing human-scene collisions withoutadditional data collection costs. Complementing this, our locomotion andhuman-scene interaction modules leverage existing methods to enrich thesystem's motion generation capabilities. Augmentation modules encompass plotcomprehension for command generation, motion synchronization for seamlessintegration of different motion types, hand pose retrieval to enhance motionrealism, motion collision revision to prevent human collisions, and 3Dretargeting to ensure visual fidelity. Experimental evaluations validate thesystem's ability to generate high-quality, diverse, and physically realisticmotions, underscoring its potential for advancing creative workflows.