Advanced LivePortrait - AI Facial Expression Editor

Real-time facial expression editing

This editor runs LivePortrait on a dedicated GPU with TensorRT and CUDA graphs, so each slider move renders a new frame in single-digit milliseconds on the GPU. Your portrait is analysed once, then every change only re-drives the face keypoints, so editing stays interactive. Uploads are held in server memory only while you edit and are not saved.

Emotion presets

Smile, grin, laugh, smirk, flirty, wink, kiss, shy, cool, proud, angry, disgust, sad, worried, pout, tired, sleepy, bored, yawn, surprised, scared, skeptical and confused are combinations of the sliders. Choose one, then scale it with Intensity or adjust any slider after.

API

Upload once to get a source id, then render as many expressions as you like. Rate limits apply per IP.

curl -s -X POST --data-binary @face.jpg -H 'Content-Type: image/jpeg' \
  'https://netwrck.com/api/liveportrait/prepare?model=human'
# {"source_id":"...","width":768,"height":1024,...}

curl -s -X POST https://netwrck.com/api/liveportrait/render \
  -H 'Content-Type: application/json' -o out.jpg \
  -d '{"source_id":"...","preset":"smirk","format":"jpeg"}'

curl -s -X POST https://netwrck.com/api/liveportrait/render \
  -H 'Content-Type: application/json' -o out.jpg \
  -d '{"source_id":"...","params":{"smile":0.8,"rotate_yaw":8,"wink":10}}'

Parameters: rotate_pitch, rotate_yaw, rotate_roll, blink, eyebrow, wink, pupil_x, pupil_y, aaa, eee, woo, smile, src_ratio. GET /api/liveportrait/schema lists ranges and presets. Send batch (up to 24 parameter sets) to get a JSON list of images in one call.