Python
{
"$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
}
}