·
4 commits
to release/EMOCA_v2
since this release
Release of EMOCA v2
New retrained of EMOCA that performs much better. Compared to the original model it produces:
- Much better lip and eye alignment
- Much better lip articulation
This is achieved by:
- Using a subset of mediapipe landmarks for mouth, eyes and eyebrows (as opposed to FAN landmarks that EMOCA v1 uses)
- Using absolute landmark loss in combination with the relative losses (as opposed to only relative landmark losses in EMOCA v1)
- Incorporating perceptual lip reading loss. Inspired by spectre. Big shout-out to these guys!
You will have to upgrade to the new environment in order to use EMOCA v2.
This includes:
- switching to Python 3.8
- updating pytorch and pytorch3d
- upating minor packages
Please follow the steps in the README