Versions 1.0
Take an image of your character, take a clip of dialog, and generate a video of that character talking, with emotions and gestures! Works for a variety of styles. Simple as cheesecake!
Inputs:
1 image of your character
1 clip of dialog, either from an audio or video file
1 prompt either simple or complex
Outputs:
1 animated video of your character talking for the length of your audio clip
Tutorial:
Cheese and have a good one!
Brie Wensleydale ~
📂 ComfyUI/
├── 📂 models/
│ ├── 📂 clip_vision/
│ │ └── clip_vision_h.safetensors
│ ├── 📂 diffusion_models/
│ │ └── Wan 2.1 I2V 14B GGUF models
│ │ └── Wan 2.1 I2V 14B FP8 models
│ ├── 📂 loras/
│ │ └── lightx2v_I2V_14B_480p_cfg_step_distill_rank64_bf16.safetensors
│ ├── 📂 vae/
│ │ └── Wan2_1_VAE_fp32.safetensors
Description
Versions 1.1
Minor tweak to the default prompt and node connections.
Previous defaults were complex and connected to silence respectively.
FAQ
Comments (4)
Fantastic. Thank you
I'm glad you found it useful!
@slipperygem One question though... regarding the "multitalk silent embeds" node, what did you mean by "run the multitalk embeds through the node"? The "multitalk silent embeds" node doesn't have any inputs so as far as I can see you can't run anything through it.
@martius72 Yeah, exactly, if you want silence and the lips remaining shut, just take the output from that node and drag it to the WanVideoSampler node.
Then it will just lip sync to silence.