- This topic has 5 replies, 4 voices, and was last updated 3 years, 3 months ago by
.
-
Posts
-
I know theres some products to collect motion captures from phone video that will convert to animated expressions, but is anyone aware of a way to use just a static image of a face to generate an expression? The video motion capture seems to only work with iphones
The way the mocap works is by having a way to map the face in 3D to project it onto the model, which is why it needs an iPhone as they have a depth sensing camera. To my knowledge there isn't a reliable way to retarget from a 2D image and get any decent level of accuracy (although with motion it can sort-of work with a webcam through virtual motion capture/other vtuber apps). You *can* capture a single frame from most of the iphone mocap apps, though.
This? https://zonegfx.com/poserecorder/
If you make a video from a face photo, around 20 seconds, perhaps work.., or in any camera, some expressions in "live".
This make a duf, and you can select the frame and "save as pose"..And if you dare, this one works better, but you need to know how to compile in Microsoft Visual Studio/Cmake, install Python (and know python...), tensoflow, etc..
https://github.com/TadasBaltrusaitis/OpenFaceThis is for MikuMikuDance (vmd files). It would be necessary to modify all the code to make a duf... LOL
https://github.com/errno-mmd/readfacevmdUsing the ControlNet module in Stable Diffusion, you can get a depth map of any image, but I do not know how to use it further.
- You must be logged in to reply to this topic.
