From 7fa989e89ca77822f39e8624c78f1859ec85a0ad Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?=E8=96=87=E6=98=9F=E5=B9=BB=E6=B5=B7?=
 <1+qishizhe@noreply.git.lain.asia>
Date: Sun, 1 Nov 2026 17:13:04 +0800
Subject: [PATCH] docs: add H3 image and reference video examples

---
 .../示例工作流/03_图生视频_H3_I2V.api.json    | 182 ++++
 .../示例工作流/03_图生视频_H3_I2V.json        | 805 ++++++++++++++++
 .../示例工作流/04_多参考对白_H3_Ref2VA.api.json | 197 ++++
 .../示例工作流/04_多参考对白_H3_Ref2VA.json   | 859 ++++++++++++++++++
 4 files changed, 2043 insertions(+)
 create mode 100644 工作流搭建指南/示例工作流/03_图生视频_H3_I2V.api.json
 create mode 100644 工作流搭建指南/示例工作流/03_图生视频_H3_I2V.json
 create mode 100644 工作流搭建指南/示例工作流/04_多参考对白_H3_Ref2VA.api.json
 create mode 100644 工作流搭建指南/示例工作流/04_多参考对白_H3_Ref2VA.json

diff --git a/工作流搭建指南/示例工作流/03_图生视频_H3_I2V.api.json b/工作流搭建指南/示例工作流/03_图生视频_H3_I2V.api.json
new file mode 100644
index 0000000..3e645cc
--- /dev/null
+++ b/工作流搭建指南/示例工作流/03_图生视频_H3_I2V.api.json
@@ -0,0 +1,182 @@
+{
+  "6": {
+    "class_type": "UNETLoader",
+    "inputs": {
+      "unet_name": "minimax_h3_fl2va_pruned_int8_convrot.safetensors",
+      "weight_dtype": "default"
+    }
+  },
+  "121": {
+    "class_type": "LoraLoaderModelOnly",
+    "inputs": {
+      "model": [
+        "6",
+        0
+      ],
+      "lora_name": "minimax_h3_fl2v_turbo_8step_v1.0_comfyui_bf16.safetensors",
+      "strength_model": 1.0
+    }
+  },
+  "13": {
+    "class_type": "CLIPLoader",
+    "inputs": {
+      "clip_name": "qwen3vl_32b_minimax_h3_nvfp4_awq.safetensors",
+      "type": "minimax",
+      "device": "default"
+    }
+  },
+  "11": {
+    "class_type": "VAELoader",
+    "inputs": {
+      "vae_name": "minimax_h3_video_vae_fp16.safetensors"
+    }
+  },
+  "24": {
+    "class_type": "VAELoader",
+    "inputs": {
+      "vae_name": "minimax_h3_audio_vae_fp32.safetensors"
+    }
+  },
+  "104": {
+    "class_type": "MiniMaxH3ImageToVideo",
+    "inputs": {
+      "clip": [
+        "13",
+        0
+      ],
+      "vae": [
+        "11",
+        0
+      ],
+      "prompt": "For the target video, at 0.00 seconds into the target video, <Picture 1> (from [Shot 1]) is fully referenced.\n\nintegrated_multimodal_description: [Shot 1] Preserve the exact reference character identities, clothes, realistic face proportions, props and light direction. Live-action movie cinematography. Slow Push In, medium close-up of the SAME elderly Moses from the reference, exact face and rust-red mantle, only a slight head motion, looks toward unseen Pharaoh just camera right. At 00:00.600 he draws a restrained breath and holds his gaze. The SAME original elderly male Mandarin voice for Moses (S1): deep warm resonant chest baritone, about 70 years old, mildly textured breath but clear strong articulation, measured grounded film acting, standard Mandarin, never a young broadcaster, never an announcer or a chant. At 00:01.300 he says with quiet authority, with a firm pause after the first word and weight on the final word: <d>[Chinese] 让我的百姓走。</d> End by 00:04.800, then his lips close, eyes hold steady and jaw tightens, no shouting and no grin. No other person speaks, no narration or extra words.\n\noverall_soundscape: A huge quiet stone interior, soft torch crackle and faint distant air, only the specified elderly Mandarin sentence with natural lip synchronization.\n\nnon_diegetic_music: None.",
+      "width": 960,
+      "height": 544,
+      "length": 175,
+      "first_frame": [
+        "1",
+        0
+      ]
+    }
+  },
+  "15": {
+    "class_type": "RandomNoise",
+    "inputs": {
+      "noise_seed": 7331420710003001
+    }
+  },
+  "16": {
+    "class_type": "BasicGuider",
+    "inputs": {
+      "model": [
+        "121",
+        0
+      ],
+      "conditioning": [
+        "104",
+        0
+      ]
+    }
+  },
+  "17": {
+    "class_type": "KSamplerSelect",
+    "inputs": {
+      "sampler_name": "res_multistep"
+    }
+  },
+  "9": {
+    "class_type": "BasicScheduler",
+    "inputs": {
+      "model": [
+        "121",
+        0
+      ],
+      "scheduler": "simple",
+      "steps": 8,
+      "denoise": 1.0
+    }
+  },
+  "14": {
+    "class_type": "SamplerCustomAdvanced",
+    "inputs": {
+      "noise": [
+        "15",
+        0
+      ],
+      "guider": [
+        "16",
+        0
+      ],
+      "sampler": [
+        "17",
+        0
+      ],
+      "sigmas": [
+        "9",
+        0
+      ],
+      "latent_image": [
+        "104",
+        1
+      ]
+    }
+  },
+  "10": {
+    "class_type": "VAEDecode",
+    "inputs": {
+      "samples": [
+        "14",
+        0
+      ],
+      "vae": [
+        "11",
+        0
+      ]
+    }
+  },
+  "23": {
+    "class_type": "VAEDecodeAudio",
+    "inputs": {
+      "samples": [
+        "14",
+        0
+      ],
+      "vae": [
+        "24",
+        0
+      ]
+    }
+  },
+  "91": {
+    "class_type": "CreateVideo",
+    "inputs": {
+      "images": [
+        "10",
+        0
+      ],
+      "audio": [
+        "23",
+        0
+      ],
+      "fps": 24.0,
+      "bit_depth": 8
+    }
+  },
+  "92": {
+    "class_type": "SaveVideo",
+    "inputs": {
+      "video": [
+        "91",
+        0
+      ],
+      "filename_prefix": "exodus_20261030/S02_demand",
+      "format": "auto",
+      "codec": "auto"
+    }
+  },
+  "1": {
+    "class_type": "LoadImage",
+    "inputs": {
+      "image": "exodus_20261030/refs/K02_demand.png"
+    }
+  }
+}
\ No newline at end of file
diff --git a/工作流搭建指南/示例工作流/03_图生视频_H3_I2V.json b/工作流搭建指南/示例工作流/03_图生视频_H3_I2V.json
new file mode 100644
index 0000000..4fca814
--- /dev/null
+++ b/工作流搭建指南/示例工作流/03_图生视频_H3_I2V.json
@@ -0,0 +1,805 @@
+{
+  "last_node_id": 121,
+  "last_link_id": 19,
+  "nodes": [
+    {
+      "id": 6,
+      "type": "UNETLoader",
+      "pos": [
+        40,
+        40
+      ],
+      "size": [
+        420,
+        320
+      ],
+      "flags": {},
+      "order": 0,
+      "mode": 0,
+      "inputs": [],
+      "outputs": [
+        {
+          "name": "MODEL",
+          "type": "MODEL",
+          "links": [
+            1
+          ]
+        }
+      ],
+      "properties": {
+        "Node name for S&R": "UNETLoader"
+      },
+      "widgets_values": [
+        "minimax_h3_fl2va_pruned_int8_convrot.safetensors",
+        "default"
+      ]
+    },
+    {
+      "id": 121,
+      "type": "LoraLoaderModelOnly",
+      "pos": [
+        580,
+        1008
+      ],
+      "size": [
+        420,
+        320
+      ],
+      "flags": {},
+      "order": 1,
+      "mode": 0,
+      "inputs": [
+        {
+          "name": "model",
+          "type": "MODEL",
+          "link": 1
+        }
+      ],
+      "outputs": [
+        {
+          "name": "MODEL",
+          "type": "MODEL",
+          "links": [
+            5,
+            7
+          ]
+        }
+      ],
+      "properties": {
+        "Node name for S&R": "LoraLoaderModelOnly"
+      },
+      "widgets_values": [
+        "minimax_h3_fl2v_turbo_8step_v1.0_comfyui_bf16.safetensors",
+        1.0
+      ]
+    },
+    {
+      "id": 13,
+      "type": "CLIPLoader",
+      "pos": [
+        40,
+        2492
+      ],
+      "size": [
+        420,
+        320
+      ],
+      "flags": {},
+      "order": 2,
+      "mode": 0,
+      "inputs": [],
+      "outputs": [
+        {
+          "name": "CLIP",
+          "type": "CLIP",
+          "links": [
+            2
+          ]
+        }
+      ],
+      "properties": {
+        "Node name for S&R": "CLIPLoader"
+      },
+      "widgets_values": [
+        "qwen3vl_32b_minimax_h3_nvfp4_awq.safetensors",
+        "minimax",
+        "default"
+      ]
+    },
+    {
+      "id": 11,
+      "type": "VAELoader",
+      "pos": [
+        40,
+        1210
+      ],
+      "size": [
+        420,
+        320
+      ],
+      "flags": {},
+      "order": 3,
+      "mode": 0,
+      "inputs": [],
+      "outputs": [
+        {
+          "name": "VAE",
+          "type": "VAE",
+          "links": [
+            3,
+            14
+          ]
+        }
+      ],
+      "properties": {
+        "Node name for S&R": "VAELoader"
+      },
+      "widgets_values": [
+        "minimax_h3_video_vae_fp16.safetensors"
+      ]
+    },
+    {
+      "id": 24,
+      "type": "VAELoader",
+      "pos": [
+        40,
+        1600
+      ],
+      "size": [
+        420,
+        320
+      ],
+      "flags": {},
+      "order": 4,
+      "mode": 0,
+      "inputs": [],
+      "outputs": [
+        {
+          "name": "VAE",
+          "type": "VAE",
+          "links": [
+            16
+          ]
+        }
+      ],
+      "properties": {
+        "Node name for S&R": "VAELoader"
+      },
+      "widgets_values": [
+        "minimax_h3_audio_vae_fp32.safetensors"
+      ]
+    },
+    {
+      "id": 104,
+      "type": "MiniMaxH3ImageToVideo",
+      "pos": [
+        580,
+        1398
+      ],
+      "size": [
+        420,
+        320
+      ],
+      "flags": {},
+      "order": 5,
+      "mode": 0,
+      "inputs": [
+        {
+          "name": "clip",
+          "type": "CLIP",
+          "link": 2
+        },
+        {
+          "name": "vae",
+          "type": "VAE",
+          "link": 3
+        },
+        {
+          "name": "first_frame",
+          "type": "IMAGE",
+          "link": 4
+        },
+        {
+          "name": "last_frame",
+          "type": "IMAGE",
+          "link": null
+        }
+      ],
+      "outputs": [
+        {
+          "name": "positive",
+          "type": "CONDITIONING",
+          "links": [
+            6
+          ]
+        },
+        {
+          "name": "LATENT",
+          "type": "LATENT",
+          "links": [
+            12
+          ]
+        }
+      ],
+      "properties": {
+        "Node name for S&R": "MiniMaxH3ImageToVideo"
+      },
+      "widgets_values": [
+        "For the target video, at 0.00 seconds into the target video, <Picture 1> (from [Shot 1]) is fully referenced.\n\nintegrated_multimodal_description: [Shot 1] Preserve the exact reference character identities, clothes, realistic face proportions, props and light direction. Live-action movie cinematography. Slow Push In, medium close-up of the SAME elderly Moses from the reference, exact face and rust-red mantle, only a slight head motion, looks toward unseen Pharaoh just camera right. At 00:00.600 he draws a restrained breath and holds his gaze. The SAME original elderly male Mandarin voice for Moses (S1): deep warm resonant chest baritone, about 70 years old, mildly textured breath but clear strong articulation, measured grounded film acting, standard Mandarin, never a young broadcaster, never an announcer or a chant. At 00:01.300 he says with quiet authority, with a firm pause after the first word and weight on the final word: <d>[Chinese] 让我的百姓走。</d> End by 00:04.800, then his lips close, eyes hold steady and jaw tightens, no shouting and no grin. No other person speaks, no narration or extra words.\n\noverall_soundscape: A huge quiet stone interior, soft torch crackle and faint distant air, only the specified elderly Mandarin sentence with natural lip synchronization.\n\nnon_diegetic_music: None.",
+        960,
+        544,
+        175
+      ]
+    },
+    {
+      "id": 15,
+      "type": "RandomNoise",
+      "pos": [
+        40,
+        430
+      ],
+      "size": [
+        420,
+        320
+      ],
+      "flags": {},
+      "order": 6,
+      "mode": 0,
+      "inputs": [],
+      "outputs": [
+        {
+          "name": "NOISE",
+          "type": "NOISE",
+          "links": [
+            8
+          ]
+        }
+      ],
+      "properties": {
+        "Node name for S&R": "RandomNoise"
+      },
+      "widgets_values": [
+        7331420710003001,
+        "fixed"
+      ]
+    },
+    {
+      "id": 16,
+      "type": "BasicGuider",
+      "pos": [
+        1120,
+        1349
+      ],
+      "size": [
+        420,
+        320
+      ],
+      "flags": {},
+      "order": 7,
+      "mode": 0,
+      "inputs": [
+        {
+          "name": "model",
+          "type": "MODEL",
+          "link": 5
+        },
+        {
+          "name": "conditioning",
+          "type": "CONDITIONING",
+          "link": 6
+        }
+      ],
+      "outputs": [
+        {
+          "name": "GUIDER",
+          "type": "GUIDER",
+          "links": [
+            9
+          ]
+        }
+      ],
+      "properties": {
+        "Node name for S&R": "BasicGuider"
+      },
+      "widgets_values": []
+    },
+    {
+      "id": 17,
+      "type": "KSamplerSelect",
+      "pos": [
+        40,
+        820
+      ],
+      "size": [
+        420,
+        320
+      ],
+      "flags": {},
+      "order": 8,
+      "mode": 0,
+      "inputs": [],
+      "outputs": [
+        {
+          "name": "SAMPLER",
+          "type": "SAMPLER",
+          "links": [
+            10
+          ]
+        }
+      ],
+      "properties": {
+        "Node name for S&R": "KSamplerSelect"
+      },
+      "widgets_values": [
+        "res_multistep"
+      ]
+    },
+    {
+      "id": 9,
+      "type": "BasicScheduler",
+      "pos": [
+        1120,
+        959
+      ],
+      "size": [
+        420,
+        320
+      ],
+      "flags": {},
+      "order": 9,
+      "mode": 0,
+      "inputs": [
+        {
+          "name": "model",
+          "type": "MODEL",
+          "link": 7
+        }
+      ],
+      "outputs": [
+        {
+          "name": "SIGMAS",
+          "type": "SIGMAS",
+          "links": [
+            11
+          ]
+        }
+      ],
+      "properties": {
+        "Node name for S&R": "BasicScheduler"
+      },
+      "widgets_values": [
+        "simple",
+        8,
+        1.0
+      ]
+    },
+    {
+      "id": 14,
+      "type": "SamplerCustomAdvanced",
+      "pos": [
+        1660,
+        1154
+      ],
+      "size": [
+        420,
+        320
+      ],
+      "flags": {},
+      "order": 10,
+      "mode": 0,
+      "inputs": [
+        {
+          "name": "noise",
+          "type": "NOISE",
+          "link": 8
+        },
+        {
+          "name": "guider",
+          "type": "GUIDER",
+          "link": 9
+        },
+        {
+          "name": "sampler",
+          "type": "SAMPLER",
+          "link": 10
+        },
+        {
+          "name": "sigmas",
+          "type": "SIGMAS",
+          "link": 11
+        },
+        {
+          "name": "latent_image",
+          "type": "LATENT",
+          "link": 12
+        }
+      ],
+      "outputs": [
+        {
+          "name": "output",
+          "type": "LATENT",
+          "links": [
+            13,
+            15
+          ]
+        },
+        {
+          "name": "denoised_output",
+          "type": "LATENT",
+          "links": []
+        }
+      ],
+      "properties": {
+        "Node name for S&R": "SamplerCustomAdvanced"
+      },
+      "widgets_values": []
+    },
+    {
+      "id": 10,
+      "type": "VAEDecode",
+      "pos": [
+        2200,
+        959
+      ],
+      "size": [
+        420,
+        320
+      ],
+      "flags": {},
+      "order": 11,
+      "mode": 0,
+      "inputs": [
+        {
+          "name": "samples",
+          "type": "LATENT",
+          "link": 13
+        },
+        {
+          "name": "vae",
+          "type": "VAE",
+          "link": 14
+        }
+      ],
+      "outputs": [
+        {
+          "name": "IMAGE",
+          "type": "IMAGE",
+          "links": [
+            17
+          ]
+        }
+      ],
+      "properties": {
+        "Node name for S&R": "VAEDecode"
+      },
+      "widgets_values": []
+    },
+    {
+      "id": 23,
+      "type": "VAEDecodeAudio",
+      "pos": [
+        2200,
+        1349
+      ],
+      "size": [
+        420,
+        320
+      ],
+      "flags": {},
+      "order": 12,
+      "mode": 0,
+      "inputs": [
+        {
+          "name": "samples",
+          "type": "LATENT",
+          "link": 15
+        },
+        {
+          "name": "vae",
+          "type": "VAE",
+          "link": 16
+        }
+      ],
+      "outputs": [
+        {
+          "name": "AUDIO",
+          "type": "AUDIO",
+          "links": [
+            18
+          ]
+        }
+      ],
+      "properties": {
+        "Node name for S&R": "VAEDecodeAudio"
+      },
+      "widgets_values": []
+    },
+    {
+      "id": 91,
+      "type": "CreateVideo",
+      "pos": [
+        2740,
+        1154
+      ],
+      "size": [
+        420,
+        320
+      ],
+      "flags": {},
+      "order": 13,
+      "mode": 0,
+      "inputs": [
+        {
+          "name": "images",
+          "type": "IMAGE",
+          "link": 17
+        },
+        {
+          "name": "audio",
+          "type": "AUDIO",
+          "link": 18
+        }
+      ],
+      "outputs": [
+        {
+          "name": "VIDEO",
+          "type": "VIDEO",
+          "links": [
+            19
+          ]
+        }
+      ],
+      "properties": {
+        "Node name for S&R": "CreateVideo"
+      },
+      "widgets_values": [
+        24.0,
+        8
+      ]
+    },
+    {
+      "id": 92,
+      "type": "SaveVideo",
+      "pos": [
+        3280,
+        1095
+      ],
+      "size": [
+        420,
+        438
+      ],
+      "flags": {},
+      "order": 14,
+      "mode": 0,
+      "inputs": [
+        {
+          "name": "video",
+          "type": "VIDEO",
+          "link": 19
+        }
+      ],
+      "outputs": [
+        {
+          "name": "video",
+          "type": "VIDEO",
+          "links": []
+        }
+      ],
+      "properties": {
+        "Node name for S&R": "SaveVideo"
+      },
+      "widgets_values": [
+        "exodus_20261030/S02_demand",
+        "auto",
+        "auto"
+      ]
+    },
+    {
+      "id": 1,
+      "type": "LoadImage",
+      "pos": [
+        40,
+        1990
+      ],
+      "size": [
+        420,
+        432
+      ],
+      "flags": {},
+      "order": 15,
+      "mode": 0,
+      "inputs": [],
+      "outputs": [
+        {
+          "name": "IMAGE",
+          "type": "IMAGE",
+          "links": [
+            4
+          ]
+        },
+        {
+          "name": "MASK",
+          "type": "MASK",
+          "links": []
+        }
+      ],
+      "properties": {
+        "Node name for S&R": "LoadImage"
+      },
+      "widgets_values": [
+        "exodus_20261030/refs/K02_demand.png",
+        "image"
+      ]
+    }
+  ],
+  "links": [
+    [
+      1,
+      6,
+      0,
+      121,
+      0,
+      "MODEL"
+    ],
+    [
+      2,
+      13,
+      0,
+      104,
+      0,
+      "CLIP"
+    ],
+    [
+      3,
+      11,
+      0,
+      104,
+      1,
+      "VAE"
+    ],
+    [
+      4,
+      1,
+      0,
+      104,
+      2,
+      "IMAGE"
+    ],
+    [
+      5,
+      121,
+      0,
+      16,
+      0,
+      "MODEL"
+    ],
+    [
+      6,
+      104,
+      0,
+      16,
+      1,
+      "CONDITIONING"
+    ],
+    [
+      7,
+      121,
+      0,
+      9,
+      0,
+      "MODEL"
+    ],
+    [
+      8,
+      15,
+      0,
+      14,
+      0,
+      "NOISE"
+    ],
+    [
+      9,
+      16,
+      0,
+      14,
+      1,
+      "GUIDER"
+    ],
+    [
+      10,
+      17,
+      0,
+      14,
+      2,
+      "SAMPLER"
+    ],
+    [
+      11,
+      9,
+      0,
+      14,
+      3,
+      "SIGMAS"
+    ],
+    [
+      12,
+      104,
+      1,
+      14,
+      4,
+      "LATENT"
+    ],
+    [
+      13,
+      14,
+      0,
+      10,
+      0,
+      "LATENT"
+    ],
+    [
+      14,
+      11,
+      0,
+      10,
+      1,
+      "VAE"
+    ],
+    [
+      15,
+      14,
+      0,
+      23,
+      0,
+      "LATENT"
+    ],
+    [
+      16,
+      24,
+      0,
+      23,
+      1,
+      "VAE"
+    ],
+    [
+      17,
+      10,
+      0,
+      91,
+      0,
+      "IMAGE"
+    ],
+    [
+      18,
+      23,
+      0,
+      91,
+      1,
+      "AUDIO"
+    ],
+    [
+      19,
+      91,
+      0,
+      92,
+      0,
+      "VIDEO"
+    ]
+  ],
+  "groups": [
+    {
+      "title": "应许之路 — S02_demand — production",
+      "bounding": [
+        -5,
+        -51,
+        3750,
+        2908
+      ],
+      "color": "#335159",
+      "font_size": 24
+    }
+  ],
+  "config": {},
+  "extra": {},
+  "version": 0.4
+}
\ No newline at end of file
diff --git a/工作流搭建指南/示例工作流/04_多参考对白_H3_Ref2VA.api.json b/工作流搭建指南/示例工作流/04_多参考对白_H3_Ref2VA.api.json
new file mode 100644
index 0000000..5ae272b
--- /dev/null
+++ b/工作流搭建指南/示例工作流/04_多参考对白_H3_Ref2VA.api.json
@@ -0,0 +1,197 @@
+{
+  "6": {
+    "class_type": "UNETLoader",
+    "inputs": {
+      "unet_name": "minimax_h3_ref2va_pruned_int8_convrot.safetensors",
+      "weight_dtype": "default"
+    }
+  },
+  "121": {
+    "class_type": "LoraLoaderModelOnly",
+    "inputs": {
+      "model": [
+        "6",
+        0
+      ],
+      "lora_name": "minimax_h3_ref2v_turbo_4step_v0.1_comfyui_bf16.safetensors",
+      "strength_model": 1.0
+    }
+  },
+  "13": {
+    "class_type": "CLIPLoader",
+    "inputs": {
+      "clip_name": "qwen3vl_32b_minimax_h3_nvfp4_awq.safetensors",
+      "type": "minimax",
+      "device": "default"
+    }
+  },
+  "11": {
+    "class_type": "VAELoader",
+    "inputs": {
+      "vae_name": "minimax_h3_video_vae_fp16.safetensors"
+    }
+  },
+  "24": {
+    "class_type": "VAELoader",
+    "inputs": {
+      "vae_name": "minimax_h3_audio_vae_fp32.safetensors"
+    }
+  },
+  "104": {
+    "class_type": "MiniMaxH3ReferenceToVideo",
+    "inputs": {
+      "clip": [
+        "13",
+        0
+      ],
+      "vae": [
+        "11",
+        0
+      ],
+      "audio_vae": [
+        "24",
+        0
+      ],
+      "prompt": "subject_definitions: <Subject 1> is Moses (S1), the elderly silver-haired Levantine man whose exact face, hair, beard, rust-red mantle, ivory tunic, straight wood staff, camera angle and lighting come from <Picture 1>, and whose ONLY speaking voice identity comes from <Audio 1>.\n\nsummary: Generate one new live-action synchronized audiovisual shot of <Subject 1> speaking the NEW Mandarin line specified below. This is identity-and-voice reference generation, not a replay of the source sentence. The target is 8.0 seconds.\n\nretention_analysis: Retain the precise recognizable face and close composition of <Picture 1>. Retain the deep mature warm textured baritone timbre, age, pronunciation and grounded dramatic style of <Audio 1>, but DO NOT repeat its words. Replace its words with the exact new <d> sentence in detailed_description. Keep the environment and lighting from <Picture 1>. Only <Subject 1> speaks. No other voice, narrator, song or subtitles.\n\ndetailed_description: [Shot 1] Preserve the exact reference character identities, clothes, realistic face proportions, props and light direction. Live-action movie cinematography. Slow Push In from the reference chest-up portrait toward the SAME exact face, no angle change. The elderly Moses looks toward frightened people just camera right, eyes briefly down as he bears their fear, then chin steadies and eyes rise. The SAME original elderly male Mandarin voice for Moses (S1): deep warm resonant chest baritone, about 70 years old, mildly textured breath but clear strong articulation, measured grounded film acting, standard Mandarin, never a young broadcaster, never an announcer or a chant. At 00:01.000 he says softly on the first phrase, then stronger without shouting: <d>[Chinese] 不要怕。站稳，看拯救临到。</d> The pause after 怕 is deliberate; end the final word by 00:06.800. His lips then close and he looks toward the sea. No extra words, no narrator, no other speaker. The voice of <Subject 1> is specifically bound to <Audio 1>, with natural synchronized mouth movement for the new Mandarin words.\n\noverall_soundscape: Storm wind and faint surf held low under the elderly voice, precise native lip synchronization, no music. The <Audio 1> reference supplies speaker identity only.\n\nnon_diegetic_music: None.",
+      "width": 960,
+      "height": 544,
+      "length": 192,
+      "ref_image_size": "max",
+      "ref_images.ref_image_1": [
+        "1",
+        0
+      ],
+      "ref_audios.ref_audio_1": [
+        "2",
+        0
+      ]
+    }
+  },
+  "15": {
+    "class_type": "RandomNoise",
+    "inputs": {
+      "noise_seed": 7331420710003005
+    }
+  },
+  "16": {
+    "class_type": "BasicGuider",
+    "inputs": {
+      "model": [
+        "121",
+        0
+      ],
+      "conditioning": [
+        "104",
+        0
+      ]
+    }
+  },
+  "17": {
+    "class_type": "KSamplerSelect",
+    "inputs": {
+      "sampler_name": "res_multistep"
+    }
+  },
+  "9": {
+    "class_type": "BasicScheduler",
+    "inputs": {
+      "model": [
+        "121",
+        0
+      ],
+      "scheduler": "simple",
+      "steps": 4,
+      "denoise": 1.0
+    }
+  },
+  "14": {
+    "class_type": "SamplerCustomAdvanced",
+    "inputs": {
+      "noise": [
+        "15",
+        0
+      ],
+      "guider": [
+        "16",
+        0
+      ],
+      "sampler": [
+        "17",
+        0
+      ],
+      "sigmas": [
+        "9",
+        0
+      ],
+      "latent_image": [
+        "104",
+        1
+      ]
+    }
+  },
+  "10": {
+    "class_type": "VAEDecode",
+    "inputs": {
+      "samples": [
+        "14",
+        0
+      ],
+      "vae": [
+        "11",
+        0
+      ]
+    }
+  },
+  "23": {
+    "class_type": "VAEDecodeAudio",
+    "inputs": {
+      "samples": [
+        "14",
+        0
+      ],
+      "vae": [
+        "24",
+        0
+      ]
+    }
+  },
+  "91": {
+    "class_type": "CreateVideo",
+    "inputs": {
+      "images": [
+        "10",
+        0
+      ],
+      "audio": [
+        "23",
+        0
+      ],
+      "fps": 24.0,
+      "bit_depth": 8
+    }
+  },
+  "92": {
+    "class_type": "SaveVideo",
+    "inputs": {
+      "video": [
+        "91",
+        0
+      ],
+      "filename_prefix": "exodus_20261030/S06_courage",
+      "format": "auto",
+      "codec": "auto"
+    }
+  },
+  "1": {
+    "class_type": "LoadImage",
+    "inputs": {
+      "image": "exodus_20261030/refs/K06_courage.png"
+    }
+  },
+  "2": {
+    "class_type": "LoadAudio",
+    "inputs": {
+      "audio": "exodus_20261030/refs/MOSES_V01_reference.wav"
+    }
+  }
+}
\ No newline at end of file
diff --git a/工作流搭建指南/示例工作流/04_多参考对白_H3_Ref2VA.json b/工作流搭建指南/示例工作流/04_多参考对白_H3_Ref2VA.json
new file mode 100644
index 0000000..3b324ad
--- /dev/null
+++ b/工作流搭建指南/示例工作流/04_多参考对白_H3_Ref2VA.json
@@ -0,0 +1,859 @@
+{
+  "last_node_id": 121,
+  "last_link_id": 21,
+  "nodes": [
+    {
+      "id": 6,
+      "type": "UNETLoader",
+      "pos": [
+        40,
+        40
+      ],
+      "size": [
+        420,
+        320
+      ],
+      "flags": {},
+      "order": 0,
+      "mode": 0,
+      "inputs": [],
+      "outputs": [
+        {
+          "name": "MODEL",
+          "type": "MODEL",
+          "links": [
+            1
+          ]
+        }
+      ],
+      "properties": {
+        "Node name for S&R": "UNETLoader"
+      },
+      "widgets_values": [
+        "minimax_h3_ref2va_pruned_int8_convrot.safetensors",
+        "default"
+      ]
+    },
+    {
+      "id": 121,
+      "type": "LoraLoaderModelOnly",
+      "pos": [
+        580,
+        1185
+      ],
+      "size": [
+        420,
+        320
+      ],
+      "flags": {},
+      "order": 1,
+      "mode": 0,
+      "inputs": [
+        {
+          "name": "model",
+          "type": "MODEL",
+          "link": 1
+        }
+      ],
+      "outputs": [
+        {
+          "name": "MODEL",
+          "type": "MODEL",
+          "links": [
+            7,
+            9
+          ]
+        }
+      ],
+      "properties": {
+        "Node name for S&R": "LoraLoaderModelOnly"
+      },
+      "widgets_values": [
+        "minimax_h3_ref2v_turbo_4step_v0.1_comfyui_bf16.safetensors",
+        1.0
+      ]
+    },
+    {
+      "id": 13,
+      "type": "CLIPLoader",
+      "pos": [
+        40,
+        2882
+      ],
+      "size": [
+        420,
+        320
+      ],
+      "flags": {},
+      "order": 2,
+      "mode": 0,
+      "inputs": [],
+      "outputs": [
+        {
+          "name": "CLIP",
+          "type": "CLIP",
+          "links": [
+            2
+          ]
+        }
+      ],
+      "properties": {
+        "Node name for S&R": "CLIPLoader"
+      },
+      "widgets_values": [
+        "qwen3vl_32b_minimax_h3_nvfp4_awq.safetensors",
+        "minimax",
+        "default"
+      ]
+    },
+    {
+      "id": 11,
+      "type": "VAELoader",
+      "pos": [
+        40,
+        1210
+      ],
+      "size": [
+        420,
+        320
+      ],
+      "flags": {},
+      "order": 3,
+      "mode": 0,
+      "inputs": [],
+      "outputs": [
+        {
+          "name": "VAE",
+          "type": "VAE",
+          "links": [
+            3,
+            16
+          ]
+        }
+      ],
+      "properties": {
+        "Node name for S&R": "VAELoader"
+      },
+      "widgets_values": [
+        "minimax_h3_video_vae_fp16.safetensors"
+      ]
+    },
+    {
+      "id": 24,
+      "type": "VAELoader",
+      "pos": [
+        40,
+        1600
+      ],
+      "size": [
+        420,
+        320
+      ],
+      "flags": {},
+      "order": 4,
+      "mode": 0,
+      "inputs": [],
+      "outputs": [
+        {
+          "name": "VAE",
+          "type": "VAE",
+          "links": [
+            4,
+            18
+          ]
+        }
+      ],
+      "properties": {
+        "Node name for S&R": "VAELoader"
+      },
+      "widgets_values": [
+        "minimax_h3_audio_vae_fp32.safetensors"
+      ]
+    },
+    {
+      "id": 104,
+      "type": "MiniMaxH3ReferenceToVideo",
+      "pos": [
+        580,
+        1575
+      ],
+      "size": [
+        420,
+        334
+      ],
+      "flags": {},
+      "order": 5,
+      "mode": 0,
+      "inputs": [
+        {
+          "name": "clip",
+          "type": "CLIP",
+          "link": 2
+        },
+        {
+          "name": "vae",
+          "type": "VAE",
+          "link": 3
+        },
+        {
+          "name": "audio_vae",
+          "type": "VAE",
+          "link": 4
+        },
+        {
+          "name": "ref_images.ref_image_1",
+          "type": "IMAGE",
+          "link": 5
+        },
+        {
+          "name": "ref_audios.ref_audio_1",
+          "type": "AUDIO",
+          "link": 6
+        }
+      ],
+      "outputs": [
+        {
+          "name": "positive",
+          "type": "CONDITIONING",
+          "links": [
+            8
+          ]
+        },
+        {
+          "name": "LATENT",
+          "type": "LATENT",
+          "links": [
+            14
+          ]
+        }
+      ],
+      "properties": {
+        "Node name for S&R": "MiniMaxH3ReferenceToVideo"
+      },
+      "widgets_values": [
+        "subject_definitions: <Subject 1> is Moses (S1), the elderly silver-haired Levantine man whose exact face, hair, beard, rust-red mantle, ivory tunic, straight wood staff, camera angle and lighting come from <Picture 1>, and whose ONLY speaking voice identity comes from <Audio 1>.\n\nsummary: Generate one new live-action synchronized audiovisual shot of <Subject 1> speaking the NEW Mandarin line specified below. This is identity-and-voice reference generation, not a replay of the source sentence. The target is 8.0 seconds.\n\nretention_analysis: Retain the precise recognizable face and close composition of <Picture 1>. Retain the deep mature warm textured baritone timbre, age, pronunciation and grounded dramatic style of <Audio 1>, but DO NOT repeat its words. Replace its words with the exact new <d> sentence in detailed_description. Keep the environment and lighting from <Picture 1>. Only <Subject 1> speaks. No other voice, narrator, song or subtitles.\n\ndetailed_description: [Shot 1] Preserve the exact reference character identities, clothes, realistic face proportions, props and light direction. Live-action movie cinematography. Slow Push In from the reference chest-up portrait toward the SAME exact face, no angle change. The elderly Moses looks toward frightened people just camera right, eyes briefly down as he bears their fear, then chin steadies and eyes rise. The SAME original elderly male Mandarin voice for Moses (S1): deep warm resonant chest baritone, about 70 years old, mildly textured breath but clear strong articulation, measured grounded film acting, standard Mandarin, never a young broadcaster, never an announcer or a chant. At 00:01.000 he says softly on the first phrase, then stronger without shouting: <d>[Chinese] 不要怕。站稳，看拯救临到。</d> The pause after 怕 is deliberate; end the final word by 00:06.800. His lips then close and he looks toward the sea. No extra words, no narrator, no other speaker. The voice of <Subject 1> is specifically bound to <Audio 1>, with natural synchronized mouth movement for the new Mandarin words.\n\noverall_soundscape: Storm wind and faint surf held low under the elderly voice, precise native lip synchronization, no music. The <Audio 1> reference supplies speaker identity only.\n\nnon_diegetic_music: None.",
+        960,
+        544,
+        192,
+        "max"
+      ]
+    },
+    {
+      "id": 15,
+      "type": "RandomNoise",
+      "pos": [
+        40,
+        430
+      ],
+      "size": [
+        420,
+        320
+      ],
+      "flags": {},
+      "order": 6,
+      "mode": 0,
+      "inputs": [],
+      "outputs": [
+        {
+          "name": "NOISE",
+          "type": "NOISE",
+          "links": [
+            10
+          ]
+        }
+      ],
+      "properties": {
+        "Node name for S&R": "RandomNoise"
+      },
+      "widgets_values": [
+        7331420710003005,
+        "fixed"
+      ]
+    },
+    {
+      "id": 16,
+      "type": "BasicGuider",
+      "pos": [
+        1120,
+        1530
+      ],
+      "size": [
+        420,
+        320
+      ],
+      "flags": {},
+      "order": 7,
+      "mode": 0,
+      "inputs": [
+        {
+          "name": "model",
+          "type": "MODEL",
+          "link": 7
+        },
+        {
+          "name": "conditioning",
+          "type": "CONDITIONING",
+          "link": 8
+        }
+      ],
+      "outputs": [
+        {
+          "name": "GUIDER",
+          "type": "GUIDER",
+          "links": [
+            11
+          ]
+        }
+      ],
+      "properties": {
+        "Node name for S&R": "BasicGuider"
+      },
+      "widgets_values": []
+    },
+    {
+      "id": 17,
+      "type": "KSamplerSelect",
+      "pos": [
+        40,
+        820
+      ],
+      "size": [
+        420,
+        320
+      ],
+      "flags": {},
+      "order": 8,
+      "mode": 0,
+      "inputs": [],
+      "outputs": [
+        {
+          "name": "SAMPLER",
+          "type": "SAMPLER",
+          "links": [
+            12
+          ]
+        }
+      ],
+      "properties": {
+        "Node name for S&R": "KSamplerSelect"
+      },
+      "widgets_values": [
+        "res_multistep"
+      ]
+    },
+    {
+      "id": 9,
+      "type": "BasicScheduler",
+      "pos": [
+        1120,
+        1140
+      ],
+      "size": [
+        420,
+        320
+      ],
+      "flags": {},
+      "order": 9,
+      "mode": 0,
+      "inputs": [
+        {
+          "name": "model",
+          "type": "MODEL",
+          "link": 9
+        }
+      ],
+      "outputs": [
+        {
+          "name": "SIGMAS",
+          "type": "SIGMAS",
+          "links": [
+            13
+          ]
+        }
+      ],
+      "properties": {
+        "Node name for S&R": "BasicScheduler"
+      },
+      "widgets_values": [
+        "simple",
+        4,
+        1.0
+      ]
+    },
+    {
+      "id": 14,
+      "type": "SamplerCustomAdvanced",
+      "pos": [
+        1660,
+        1335
+      ],
+      "size": [
+        420,
+        320
+      ],
+      "flags": {},
+      "order": 10,
+      "mode": 0,
+      "inputs": [
+        {
+          "name": "noise",
+          "type": "NOISE",
+          "link": 10
+        },
+        {
+          "name": "guider",
+          "type": "GUIDER",
+          "link": 11
+        },
+        {
+          "name": "sampler",
+          "type": "SAMPLER",
+          "link": 12
+        },
+        {
+          "name": "sigmas",
+          "type": "SIGMAS",
+          "link": 13
+        },
+        {
+          "name": "latent_image",
+          "type": "LATENT",
+          "link": 14
+        }
+      ],
+      "outputs": [
+        {
+          "name": "output",
+          "type": "LATENT",
+          "links": [
+            15,
+            17
+          ]
+        },
+        {
+          "name": "denoised_output",
+          "type": "LATENT",
+          "links": []
+        }
+      ],
+      "properties": {
+        "Node name for S&R": "SamplerCustomAdvanced"
+      },
+      "widgets_values": []
+    },
+    {
+      "id": 10,
+      "type": "VAEDecode",
+      "pos": [
+        2200,
+        1140
+      ],
+      "size": [
+        420,
+        320
+      ],
+      "flags": {},
+      "order": 11,
+      "mode": 0,
+      "inputs": [
+        {
+          "name": "samples",
+          "type": "LATENT",
+          "link": 15
+        },
+        {
+          "name": "vae",
+          "type": "VAE",
+          "link": 16
+        }
+      ],
+      "outputs": [
+        {
+          "name": "IMAGE",
+          "type": "IMAGE",
+          "links": [
+            19
+          ]
+        }
+      ],
+      "properties": {
+        "Node name for S&R": "VAEDecode"
+      },
+      "widgets_values": []
+    },
+    {
+      "id": 23,
+      "type": "VAEDecodeAudio",
+      "pos": [
+        2200,
+        1530
+      ],
+      "size": [
+        420,
+        320
+      ],
+      "flags": {},
+      "order": 12,
+      "mode": 0,
+      "inputs": [
+        {
+          "name": "samples",
+          "type": "LATENT",
+          "link": 17
+        },
+        {
+          "name": "vae",
+          "type": "VAE",
+          "link": 18
+        }
+      ],
+      "outputs": [
+        {
+          "name": "AUDIO",
+          "type": "AUDIO",
+          "links": [
+            20
+          ]
+        }
+      ],
+      "properties": {
+        "Node name for S&R": "VAEDecodeAudio"
+      },
+      "widgets_values": []
+    },
+    {
+      "id": 91,
+      "type": "CreateVideo",
+      "pos": [
+        2740,
+        1335
+      ],
+      "size": [
+        420,
+        320
+      ],
+      "flags": {},
+      "order": 13,
+      "mode": 0,
+      "inputs": [
+        {
+          "name": "images",
+          "type": "IMAGE",
+          "link": 19
+        },
+        {
+          "name": "audio",
+          "type": "AUDIO",
+          "link": 20
+        }
+      ],
+      "outputs": [
+        {
+          "name": "VIDEO",
+          "type": "VIDEO",
+          "links": [
+            21
+          ]
+        }
+      ],
+      "properties": {
+        "Node name for S&R": "CreateVideo"
+      },
+      "widgets_values": [
+        24.0,
+        8
+      ]
+    },
+    {
+      "id": 92,
+      "type": "SaveVideo",
+      "pos": [
+        3280,
+        1276
+      ],
+      "size": [
+        420,
+        438
+      ],
+      "flags": {},
+      "order": 14,
+      "mode": 0,
+      "inputs": [
+        {
+          "name": "video",
+          "type": "VIDEO",
+          "link": 21
+        }
+      ],
+      "outputs": [
+        {
+          "name": "video",
+          "type": "VIDEO",
+          "links": []
+        }
+      ],
+      "properties": {
+        "Node name for S&R": "SaveVideo"
+      },
+      "widgets_values": [
+        "exodus_20261030/S06_courage",
+        "auto",
+        "auto"
+      ]
+    },
+    {
+      "id": 1,
+      "type": "LoadImage",
+      "pos": [
+        40,
+        1990
+      ],
+      "size": [
+        420,
+        432
+      ],
+      "flags": {},
+      "order": 15,
+      "mode": 0,
+      "inputs": [],
+      "outputs": [
+        {
+          "name": "IMAGE",
+          "type": "IMAGE",
+          "links": [
+            5
+          ]
+        },
+        {
+          "name": "MASK",
+          "type": "MASK",
+          "links": []
+        }
+      ],
+      "properties": {
+        "Node name for S&R": "LoadImage"
+      },
+      "widgets_values": [
+        "exodus_20261030/refs/K06_courage.png",
+        "image"
+      ]
+    },
+    {
+      "id": 2,
+      "type": "LoadAudio",
+      "pos": [
+        40,
+        2492
+      ],
+      "size": [
+        420,
+        320
+      ],
+      "flags": {},
+      "order": 16,
+      "mode": 0,
+      "inputs": [],
+      "outputs": [
+        {
+          "name": "AUDIO",
+          "type": "AUDIO",
+          "links": [
+            6
+          ]
+        }
+      ],
+      "properties": {
+        "Node name for S&R": "LoadAudio"
+      },
+      "widgets_values": [
+        "exodus_20261030/refs/MOSES_V01_reference.wav"
+      ]
+    }
+  ],
+  "links": [
+    [
+      1,
+      6,
+      0,
+      121,
+      0,
+      "MODEL"
+    ],
+    [
+      2,
+      13,
+      0,
+      104,
+      0,
+      "CLIP"
+    ],
+    [
+      3,
+      11,
+      0,
+      104,
+      1,
+      "VAE"
+    ],
+    [
+      4,
+      24,
+      0,
+      104,
+      2,
+      "VAE"
+    ],
+    [
+      5,
+      1,
+      0,
+      104,
+      3,
+      "IMAGE"
+    ],
+    [
+      6,
+      2,
+      0,
+      104,
+      4,
+      "AUDIO"
+    ],
+    [
+      7,
+      121,
+      0,
+      16,
+      0,
+      "MODEL"
+    ],
+    [
+      8,
+      104,
+      0,
+      16,
+      1,
+      "CONDITIONING"
+    ],
+    [
+      9,
+      121,
+      0,
+      9,
+      0,
+      "MODEL"
+    ],
+    [
+      10,
+      15,
+      0,
+      14,
+      0,
+      "NOISE"
+    ],
+    [
+      11,
+      16,
+      0,
+      14,
+      1,
+      "GUIDER"
+    ],
+    [
+      12,
+      17,
+      0,
+      14,
+      2,
+      "SAMPLER"
+    ],
+    [
+      13,
+      9,
+      0,
+      14,
+      3,
+      "SIGMAS"
+    ],
+    [
+      14,
+      104,
+      1,
+      14,
+      4,
+      "LATENT"
+    ],
+    [
+      15,
+      14,
+      0,
+      10,
+      0,
+      "LATENT"
+    ],
+    [
+      16,
+      11,
+      0,
+      10,
+      1,
+      "VAE"
+    ],
+    [
+      17,
+      14,
+      0,
+      23,
+      0,
+      "LATENT"
+    ],
+    [
+      18,
+      24,
+      0,
+      23,
+      1,
+      "VAE"
+    ],
+    [
+      19,
+      10,
+      0,
+      91,
+      0,
+      "IMAGE"
+    ],
+    [
+      20,
+      23,
+      0,
+      91,
+      1,
+      "AUDIO"
+    ],
+    [
+      21,
+      91,
+      0,
+      92,
+      0,
+      "VIDEO"
+    ]
+  ],
+  "groups": [
+    {
+      "title": "应许之路 — S06_courage — production",
+      "bounding": [
+        -5,
+        -51,
+        3750,
+        3298
+      ],
+      "color": "#335159",
+      "font_size": 24
+    }
+  ],
+  "config": {},
+  "extra": {},
+  "version": 0.4
+}
\ No newline at end of file