Replies: 1 comment
-
The Pose model is meant for tracking certain landmarks that can then be used to control bones. Face mesh is specifically designed to find the contours of the face. Pose doesn't work that way. https://google.github.io/mediapipe/solutions/pose.html What you need is a skinned mesh that Pose can control. Then you can texture that model the way you did the face. You can probably get started with something from https://www.mixamo.com/ |
Beta Was this translation helpful? Give feedback.
0 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Uh oh!
There was an error while loading. Please reload this page.
-
I'd like to create something similar to this Mediapipe FaceMesh example, but applying texture on the body instead of the face. Mediapipe Body doesn't provide as many coordinates as Facemesh so I'm having a hard time coming up with the appropriate approach for this.
With Facemesh I could loop through the face coordinates and create an array of Three.Vector objs.
faceCoords = Array.from(faceCoords,x=>new THREE.Vector3(x.x,x.y,x.z));
facialMesh = new THREE.Mesh(dotGeometry, Materials.meshMaterial)
I loop through the faceCoords array and map each position in facialMesh to the coordinates in faceCoords.
How can I apply a similar technique using Mediapipe Pose when it only provides 33 points as opposed to the 468 Facemesh provides?
Beta Was this translation helpful? Give feedback.
All reactions