Computational storytelling systems have mainly focused on the construction and evaluation of textual discourse for communicating stories. Few intelligentcamera systems have been built in 3D environments for effective...
详细信息
ISBN:
(纸本)9783642106422
Computational storytelling systems have mainly focused on the construction and evaluation of textual discourse for communicating stories. Few intelligentcamera systems have been built in 3D environments for effective visual communication of stories. The evaluation of effectiveness of these systems, if any, has focused mainly on the run-time performance of the camera placement algorithms. The purpose of this paper is to present a systematic cognitive-based evaluation methodology to compare effects of different cinematic visualization strategies on viewer comprehension of stories. In particular, an evaluation of automatically generated visualizations from Darshak, a cinematic planning system, against different hand-generated visualization strategies is presented. The methodology used in the empirical evaluation is based on QUEST, a cognitive framework for question-answering in the context of stories, that provides validated predictors for measuring story coherence in readers. Data collected from viewers, who watch the same story renedered with three different visualization strategies, is compared with QUEST's predictor metrics. Initial data analysis establishes significant effect on choice of visualization strategy on story comprehension. It further shows a significant effect of visualization strategy selected by Darshak on viewers' measured story coherence.
In this work. we have developed a real-time cameracontrol module for navigation in virtual environments. With this module, the tracking motion of a third-person camera can be generated automatically to allow a user t...
详细信息
ISBN:
(纸本)9783540854104
In this work. we have developed a real-time cameracontrol module for navigation in virtual environments. With this module, the tracking motion of a third-person camera can be generated automatically to allow a user to focus on the control of an avatar. The core of this module consists of a motion planner that uses the probabilistic roadmap method and a lazy update strategy to generate the motion of the camera, possibly with necessary intercuts. A dynamic roadmap specified relative to the avatar is updated in real time within a time budget to account for occlusions in every frame of the control loop. In addition, the planner also allows a user to specify preferences on how the tracking motion is generated. We will use several examples to demonstrate the effectiveness of this real-time camera planning system.
暂无评论