Skip to content
Python
5.0 KB
Raw Blame
{
  "$schema": "../../schema/app.schema.json",
  "id": "talking",
  "name": "Talking Avatar",
  "section": "video",
  "pack": "video-ltx25",
  "status": "ready",
  "higgsfield_feature": "Lipsync / Speak",
  "summary": "A portrait speaks or sings along to your audio, with LTX-2.5 audio-to-video in two stages.",
  "workflow": "workflow.api.json",
  "source": {
    "template": "ComfyUI-LTXVideo example_workflows/2.5/LTX-2.5_A2V_Two_Stage_Distilled.json (44c8a36)",
    "comfyui": "0.42.0",
    "notes": "Gemma API text encoder removed (local encoding only), int8 weights, portrait conditioning always on. Not run yet."
  },
  "inputs": [
    {
      "id": "portrait",
      "type": "image",
      "label": "Portrait",
      "required": true,
      "limits": {
        "max_mb": 20
      },
      "bind": [
        {
          "node": "2004",
          "field": "image"
        }
      ]
    },
    {
      "id": "audio",
      "type": "audio",
      "label": "Voice or song",
      "required": true,
      "limits": {
        "max_mb": 50,
        "seconds": [
          2,
          20
        ]
      },
      "bind": [
        {
          "node": "5600",
          "field": "audio"
        }
      ]
    },
    {
      "id": "prompt",
      "type": "textarea",
      "label": "Describe the performance",
      "required": true,
      "default": "She speaks warmly to the camera with natural head movement and expressive eyes, soft studio light",
      "bind": [
        {
          "node": "5508",
          "field": "value"
        }
      ]
    },
    {
      "id": "negative",
      "type": "textarea",
      "label": "Must not appear",
      "default": "blurry, distorted face, warped mouth",
      "bind": [
        {
          "node": "5509",
          "field": "value"
        }
      ]
    },
    {
      "id": "start",
      "type": "number",
      "label": "Audio start (seconds)",
      "min": 0,
      "max": 600,
      "default": 0,
      "bind": [
        {
          "node": "5601",
          "field": "value"
        }
      ]
    },
    {
      "id": "duration",
      "type": "slider",
      "label": "Seconds",
      "min": 2,
      "max": 10,
      "step": 1,
      "default": 5,
      "bind": [
        {
          "node": "5512",
          "field": "value"
        }
      ]
    },
    {
      "id": "enhance",
      "type": "toggle",
      "label": "Enhance prompt",
      "default": false,
      "help": "Rewrites your prompt with a small Gemma 4 model that also looks at the portrait",
      "bind": [
        {
          "node": "5014:5556",
          "field": "switch"
        }
      ]
    },
    {
      "id": "seed",
      "type": "seed",
      "label": "Seed",
      "default": -1,
      "help": "-1 picks a random seed",
      "bind": [
        {
          "node": "5516:4832",
          "field": "noise_seed"
        }
      ]
    }
  ],
  "outputs": [
    {
      "node": "4852",
      "type": "video",
      "label": "Video"
    }
  ],
  "models": [
    {
      "repo": "Lightricks/LTX-2.5",
      "file": "diffusion_models/ltx-2.5-22b-distilled-transformer-comfy-int8-convrot.safetensors",
      "dir": "diffusion_models",
      "gb": 21.5,
      "phase": "denoise",
      "gated": true,
      "license": "LTX-2.x Community License"
    },
    {
      "repo": "Lightricks/LTX-2.5",
      "file": "text_encoders/gemma4-12b-with-proj-ltx-2.5-comfy-int8-convrot.safetensors",
      "dir": "text_encoders",
      "gb": 15.37,
      "phase": "encode",
      "gated": true,
      "license": "LTX-2.x Community License"
    },
    {
      "repo": "Lightricks/LTX-2.5",
      "file": "vae/ltx-2.5-video-vae-bf16.safetensors",
      "dir": "vae",
      "gb": 1.47,
      "phase": "decode",
      "gated": true,
      "license": "LTX-2.x Community License"
    },
    {
      "repo": "Lightricks/LTX-2.5",
      "file": "vae/ltx-2.5-audio-vae-bf16.safetensors",
      "dir": "vae",
      "gb": 0.36,
      "phase": "decode",
      "gated": true,
      "license": "LTX-2.x Community License"
    },
    {
      "repo": "Lightricks/LTX-2.5",
      "file": "latent_upscale_models/ltx-2.5-latent-spatial-upscaler-x2-bf16-1.0.safetensors",
      "dir": "latent_upscale_models",
      "gb": 1.0,
      "phase": "aux",
      "gated": true,
      "license": "LTX-2.x Community License"
    },
    {
      "repo": "Comfy-Org/gemma-4",
      "file": "text_encoders/gemma4_e2b_it_int8_convrot.safetensors",
      "dir": "text_encoders",
      "gb": 5.2,
      "phase": "encode",
      "license": "Apache-2.0",
      "optional": true
    },
    {
      "repo": "elix3r/LTX-2.3-22b-AV-LoRA-talking-head",
      "file": "LTX-2.3-22b-AV-LoRA-talking-head-v1.safetensors",
      "dir": "loras",
      "gb": 0.43,
      "phase": "denoise",
      "license": "OpenRAIL",
      "optional": true
    }
  ],
  "notes": "Use footage and photos of yourself, fictional characters, or people who agreed to it.",
  "credits": [
    "LTX-2.5 by Lightricks",
    "Workflow adapted from Lightricks' ComfyUI-LTXVideo examples"
  ],
  "test": {
    "portrait": "test/portrait.png",
    "audio": "test/voice.wav",
    "duration": 2
  }
}