International Research Journal of Engineering and Technology (IRJET) Volume: 09 Issue: 06 | June 2022
www.irjet.net
e-ISSN: 2395-0056 p-ISSN: 2395-0072
Motion capture for Animation Naman Luniya1, Rishav2, Ronak Jain3 Department of Computer Science Engineering, MIT School of Engineering,MIT-ADT University, Pune, Maharashtra, INDIA ------------------------------------------------------------------------***-------------------------------------------------------------------Abstract 1,2,3
Artificial intelligence and machine learning are becoming increasingly popular. In the realm of performance capture for the entertainment sector, we are investigating the use of artificial intelligence and machine learning. The present methods are generally prohibitively expensive for the general public. We are concentrating on using artificial intelligence and external software (such as Unity) together to create amazing motion capture outcomes. This will lower the cost of inputs while also increasing innovation.
Keywords – Performance capture, Motioncapture, AI/ML, Unity. 1. Introduction The expansion of technological impact in the entertainment sector has shown to be effective. Motion Capture (MoCap) is the process of using cameras and specifically built suits to capture the movement of objects or people. Its impact on the entertainment business has been significant. Motion capture has been popular in movies and video games since the 1970s. In comparison to previous procedures, it has been complimented for its reduced latency, replication of complex moves, and data production. The demand for specific software and hardware, as well as the expense of relevant inputs, has been widely criticized. The goal of this project is to drastically lower the cost of MoCap. The app eliminates the need for sophisticated lenses and motion capture suits. With the help of Python's OpenCV, scripting and Unity, we were able tocomplete the job. The user uploads video to the app, which is then pre-processed with OpenCV & Unity, which aids in the creation of animated clips of the user.These procedures demonstrate the capacity to capture performance in a coherent way.
2. Literature Survey There has been work conducted in the field of face detection & its recognition along with development of mocap suits for better performance capture. The previous works have been focused on for better detection and capturing of coordinates. The prominent methods for face detection in this field are Eigen face detection, ADABOOST algorithm, Convex hull algorithm and Voila & Jones face detection. The development in capturing of coordinates will help to create animations easier than to follow the tedious & expensive method of mocap suits. The process of face detection is done using a single algorithm or with a combination of algorithms. Eigen Face detection is used with a combination of other algorithms[2]. It lacks in performance in comparison with the Haar Cascade classifier. The Haar cascade classifier is the most efficient in terms of face detection and recognition [2,4].The downside being its sensitivity to the background. The ADABOOST algorithm is efficient in capturing coordinates for the detection, especially for hand and fingers [4]. It has shown better speed and accuracy with a drawback for longer trainingperiods[4,13]. Motion capture has been prominently done usingthe motion capture suits. The suits are expensive for the general populace [8]. The work-faced issues related to capturing of limbs, legs & coordinates. Studies conducted on its cost effectiveness along with maintaining its standardstill lack solidity. Thus, there is a need to find alternate pathways for motion capturing.
© 2022, IRJET
|
Impact Factor value: 7.529
|
ISO 9001:2008 Certified Journal
|
Page 2335