Single ComfyUI node that auto-scales Wan2.1 InfiniTalk video generation to any audio duration. No manual segment chaining needed. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
130 lines
3.2 KiB
JSON
130 lines
3.2 KiB
JSON
{
|
|
"unet": {
|
|
"inputs": {
|
|
"unet_name": "Wan2_1-I2V-14B-480p_fp8_e4m3fn_scaled_KJ.safetensors",
|
|
"weight_dtype": "default"
|
|
},
|
|
"class_type": "UNETLoader",
|
|
"_meta": { "title": "Load Diffusion Model" }
|
|
},
|
|
"clip": {
|
|
"inputs": {
|
|
"clip_name": "umt5_xxl_fp8_e4m3fn_scaled.safetensors",
|
|
"type": "wan",
|
|
"device": "default"
|
|
},
|
|
"class_type": "CLIPLoader",
|
|
"_meta": { "title": "Load CLIP" }
|
|
},
|
|
"vae": {
|
|
"inputs": {
|
|
"vae_name": "Wan2_1_VAE_bf16.safetensors"
|
|
},
|
|
"class_type": "VAELoader",
|
|
"_meta": { "title": "Load VAE" }
|
|
},
|
|
"lora": {
|
|
"inputs": {
|
|
"lora_name": "lightx2v_I2V_14B_480p_cfg_step_distill_rank64_bf16.safetensors",
|
|
"strength_model": 1,
|
|
"model": ["unet", 0]
|
|
},
|
|
"class_type": "LoraLoaderModelOnly",
|
|
"_meta": { "title": "LightX2V Distill LoRA" }
|
|
},
|
|
"patch": {
|
|
"inputs": {
|
|
"name": "wan2.1_infiniteTalk_multi_fp16.safetensors"
|
|
},
|
|
"class_type": "ModelPatchLoader",
|
|
"_meta": { "title": "InfiniTalk Model Patch" }
|
|
},
|
|
"audio_enc": {
|
|
"inputs": {
|
|
"audio_encoder_name": "wav2vec2-chinese-base_fp16.safetensors"
|
|
},
|
|
"class_type": "AudioEncoderLoader",
|
|
"_meta": { "title": "Wav2Vec Audio Encoder" }
|
|
},
|
|
"prompt": {
|
|
"inputs": {
|
|
"text": "A person speaking naturally to camera, medium close-up, soft lighting, shallow depth of field",
|
|
"clip": ["clip", 0]
|
|
},
|
|
"class_type": "CLIPTextEncode",
|
|
"_meta": { "title": "Prompt" }
|
|
},
|
|
"neg": {
|
|
"inputs": {
|
|
"conditioning": ["prompt", 0]
|
|
},
|
|
"class_type": "ConditioningZeroOut",
|
|
"_meta": { "title": "Negative (Zero)" }
|
|
},
|
|
"image": {
|
|
"inputs": {
|
|
"image": "your_face_image.png"
|
|
},
|
|
"class_type": "LoadImage",
|
|
"_meta": { "title": "Reference Face Image" }
|
|
},
|
|
"audio": {
|
|
"inputs": {
|
|
"audio": "your_voiceover.mp3"
|
|
},
|
|
"class_type": "LoadAudio",
|
|
"_meta": { "title": "Voiceover Audio" }
|
|
},
|
|
"audio_emb": {
|
|
"inputs": {
|
|
"audio_encoder": ["audio_enc", 0],
|
|
"audio": ["audio", 0]
|
|
},
|
|
"class_type": "AudioEncoderEncode",
|
|
"_meta": { "title": "Encode Audio" }
|
|
},
|
|
"autoscale": {
|
|
"inputs": {
|
|
"model": ["lora", 0],
|
|
"model_patch": ["patch", 0],
|
|
"positive": ["prompt", 0],
|
|
"negative": ["neg", 0],
|
|
"vae": ["vae", 0],
|
|
"audio_encoder_output_1": ["audio_emb", 0],
|
|
"start_image": ["image", 0],
|
|
"width": 832,
|
|
"height": 480,
|
|
"frame_window": 81,
|
|
"motion_frame_count": 9,
|
|
"max_segments": 0,
|
|
"steps": 4,
|
|
"cfg": 1.0,
|
|
"sampler_name": "euler",
|
|
"scheduler": "normal",
|
|
"seed": 42,
|
|
"audio_scale": 1.0
|
|
},
|
|
"class_type": "InfiniTalkAutoScale",
|
|
"_meta": { "title": "InfiniTalk AutoScale" }
|
|
},
|
|
"video": {
|
|
"inputs": {
|
|
"fps": 25,
|
|
"images": ["autoscale", 0],
|
|
"audio": ["audio", 0]
|
|
},
|
|
"class_type": "CreateVideo",
|
|
"_meta": { "title": "Create Video" }
|
|
},
|
|
"save": {
|
|
"inputs": {
|
|
"filename_prefix": "video/infinitalk_autoscale",
|
|
"format": "auto",
|
|
"codec": "auto",
|
|
"video": ["video", 0]
|
|
},
|
|
"class_type": "SaveVideo",
|
|
"_meta": { "title": "Save Video" }
|
|
}
|
|
}
|