search

YuanxunLu / LiveSpeechPortraits

Live Speech Portraits: Real-Time Photorealistic Talking-Head Animation (SIGGRAPH Asia 2021)
MIT License
1.16k stars 198 forks source link

how to transform 2d shoulder points to 3d shoulder points? #66

Open foocker opened 2 years ago

foocker commented 2 years ago

as see. i have search some resources, but it is still a problem for me at present. thx, as u say, "by assuming a billboard model where the depth is the mean 3D facial depth for each frame." there only a little infomation about billboard model, wich make me still confusion...

YuanxunLu commented 2 years ago

Check the code and the citing paper.

foocker commented 2 years ago

Check the code and the citing paper.

for any camera model, we must know the depth of every z value of 2Dshoulder. from the paper, billboard model like a canvan, the depth is the mean of z on all traing data. but the given 3Dshoulder.npy ,z value from which is not equal for every different frame. and i will read the citing paper...