Face Motion Capture and whole body motion capture are used a lot, but do you know what methods have been adopted for the technical implementation of motion capture? Let’s take a look together.
1. Marked points
Facial motion capture systems based on marking points are more common. The number of marking points is uncertain, which is determined by the supporting equipment and systems. The number of facial markers may reach up to 350. The marking points can be drawn or pasted, and need to be used in conjunction with high-resolution cameras and other equipment. This operation is relatively cumbersome.
devices are usually head-mounted. These devices can be used in conjunction with the posture capture system to keep the actor's performance process coherent and unaffected. Depending on the shooting scene and implementation method, sometimes it will be replaced by others to help hand-held shooting.
2, Optical Motion Capture
Optical Motion Capture is to track the marks on the optical motion capture suit, synchronize the marking data under cameras at different perspectives, and use a three-dimensional reconstruction algorithm to reconstruct the motion data of different parts of the human body.
3, Inertial Motion Capture
Inertial Motion Capture is to record the data of the inertial sensor on the wearable device to obtain motion data, and finally reconstruct the three-dimensional motion trajectory through software and convert it into skeletal animation, thereby driving the virtual character.
4, markless
0, markless
, markless lattice methods usually rely on iconic positions such as nostrils, corners of the eyes, lips, dimples, etc. to determine the expression and movement status of the face. This method was first implemented by CMU, IBM, University of Manchester and other institutions through the use of subjective expression models, principal component analysis and other models and technologies.
5, AI video motion capture
AI video motion capture is to extract multiple single-frame images from the video, and use AI algorithm to extract human skeleton poses from different frames of images, and connect human skeleton poses in chronological order to form skeleton animation data, which can be used to drive virtual characters.
Do you know all the motion capture techniques?
