延迟(音频)
触地
出钢
计算机科学
运动(音乐)
实时计算
人工智能
计算机视觉
工程类
声学
电信
物理
机械工程
考古
历史
作者
Haijun Xia,Ricardo Jota,Benjamin McCanny,Zhe Yu,Clifton Forlines,Karan Singh,Daniel Wigdor
标识
DOI:10.1145/2642918.2647348
摘要
A method of reducing the perceived latency of touch input by employing a model to predict touch events before the finger reaches the touch surface is proposed. A corpus of 3D finger movement data was collected, and used to develop a model capable of three granularities at different phases of movement: initial direction, final touch location, time of touchdown. The model is validated for target distances >= 25.5cm, and demonstrated to have a mean accuracy of 1.05cm 128ms before the user touches the screen. Preference study of different levels of latency reveals a strong preference for unperceived latency touchdown feedback. A form of 'soft' feedback, as well as other uses for this prediction to improve performance, is proposed.
科研通智能强力驱动
Strongly Powered by AbleSci AI