workflow_api_ref2va_turbo_keyframe.json

{
 "1": {
  "class_type": "UNETLoader",
  "inputs": {
   "unet_name": "MiniMaxH3\\MiniMax_H3_Ref2VA_pruned_nvfp4.safetensors",
   "weight_dtype": "default"
  }
 },
 "2": {
  "class_type": "CLIPLoader",
  "inputs": {
   "clip_name": "MiniMaxH3\\qwen3vl_32b_minimax_h3_int4_convrot.safetensors",
   "type": "minimax",
   "device": "default"
  }
 },
 "3": {
  "class_type": "VAELoader",
  "inputs": {
   "vae_name": "MiniMaxH3\\minimax_h3_video_vae_fp16.safetensors"
  }
 },
 "4": {
  "class_type": "VAELoader",
  "inputs": {
   "vae_name": "MiniMaxH3\\minimax_h3_audio_vae_fp32.safetensors"
  }
 },
 "20": {
  "class_type": "MiniMaxH3TurboLoRA",
  "inputs": {
   "model": [
    "1",
    0
   ],
   "lora_name": "MiniMaxH3\\minimax_h3_turbo_v4_step600_ema.safetensors",
   "strength": 1.0,
   "low_vram": false
  }
 },
 "8": {
  "class_type": "MiniMaxH3ReferenceToVideo",
  "inputs": {
   "clip": [
    "2",
    0
   ],
   "vae": [
    "3",
    0
   ],
   "audio_vae": [
    "4",
    0
   ],
   "prompt": "subject_definitions:\n<Subject 1> is the man in <Picture 1> and <Picture 2>, around 45 years old, with wavy dark-brown hair with a slight quiff, a full salt-and-pepper beard, wearing a burgundy blazer over a cream turtleneck sweater.\n<Subject 2> is the old man in <Picture 3>, about seventy years old, bald on top with a fringe of white hair around the sides, a broad snub nose, wide-set protruding eyes and a thick white beard, wearing a plain undyed wool himation over a coarse chiton, barefoot.\n<Subject 3> is an Athenian of about fifty, tall and lean, clean-shaven, with close-cropped grey hair receding at the temples, a long straight nose and a heavy jaw, wearing a good himation dyed deep blue over a pale chiton and leather sandals.\n<Video 1> is the source video that the target video continues from, and its final frame is the starting point of the new shot.\n\nsummary:\n[video continuation + reference generation] The target video continues the shot in <Video 1>: the man with the coins leaves and the camera comes round onto <Subject 2>, alone, barefoot and empty-handed.\n\nretention_analysis:\n<Subject 1> (already present in <Video 1>): fully_preserved - the man is ALREADY IN THE INHERITED FRAME and simply continues to exist there; he is not added to the shot and he is not instantiated a second time. Exactly one instance of him exists in the video, and on that one instance his identity, wavy dark-brown hair, full salt-and-pepper beard, burgundy blazer and cream turtleneck are retained.\n<Subject 2> (appears in [Shot 1]): fully_preserved - the old man's bald crown with the fringe of white hair, broad snub nose, thick white beard and plain undyed himation are retained.\n<Video 1>: continued - the target video resumes from its final frame, keeping the same set, lighting, lens and camera movement.\n\ndetailed_description:\nThe target video is in a photorealistic cinematic documentary style with natural skin texture and shallow depth of field.\n[Shot 1] The shot continues seamlessly from the final frame of <Video 1>, with no cut between them: the same corner of the agora, the same low warm light from the left, the same lens. The camera is mounted on a dolly and its movement is perfectly steady from the first frame to the last: it is never handheld, and there is no shake, no vibration, no jitter, no sudden reframing and no zoom at any point. The arc continues in the same direction and at the same speed, leaving the man in the burgundy blazer and coming round and down onto the old man alone, ending on a medium shot that holds him from the knees up, so that the first and the last frame of the shot hold entirely different subjects. The camera then holds that framing without moving again. <Subject 3> (S3), the tall lean clean-shaven Athenian of about fifty with close-cropped grey hair and a long straight nose in a deep-blue himation, closes his hand over the bronze coins, lowers it and walks out of frame to the left. <Subject 2> (S2), the old man with the bald crown and the fringe of white hair, the broad snub nose and the thick white beard, is left standing alone on the worn stones: his undyed himation is thin and frayed at the hem and pulled around him in one plain fold, his feet are bare and dusty, his hands hang open and empty at his sides, and he watches the man go without moving after him. His lips stay closed and he does not speak. The emptying square behind him is kept softly out of focus. <Subject 1> (S1), the man with wavy dark-brown hair with a slight quiff, a full salt-and-pepper beard, a burgundy blazer and a cream turtleneck, stands at the right edge of the frame in the shade of a column, well out of focus and turned three-quarters away; he does not speak and his mouth stays closed for the whole shot. The light, the colour and the exposure do not change during the shot. There are no metal fences, no metal railings, no wire mesh, no grilles and no manufactured object of any later age anywhere in the frame; no plastic, no rubber, no synthetic sheeting, no bubbled surfaces and no net-like weave on any wall, floor or sky. Every surface is dry and worked: pitted limestone, sun-bleached wood, undyed wool, darkened beeswax, dull bronze, fired clay. There is exactly one man in a burgundy blazer and cream turtleneck in the whole frame, at every moment of the shot and at every distance from the lens: no second man dressed like him, no twin, no double and no lookalike appears anywhere, neither in focus nor out of focus, neither in the foreground nor among the people in the background. Every other person in the shot is a distinct ancient Greek in undyed or dyed wool, each with a clearly different face, different hair and a different build from <Subject 1>, and none of them has his hairline, his wavy dark-brown hair, his salt-and-pepper beard or his features; no ancient Greek in the frame wears his face. In the last third of the shot the framing holds on the old man alone, nobody comes back to him, and no new face appears in the frame from that moment to the end.\n\noverall_soundscape:\nThe edge of the square at dusk: coins closed into a fist and carried away, bare feet on worn stone, cloth dragging, the market packing up far off. No modern sounds.\n\nnon_diegetic_music:\nN/A",
   "width": 1344,
   "height": 768,
   "length": 107,
   "ref_image_size": "match",
   "ref_images.ref_image_0": [
    "100",
    0
   ],
   "ref_images.ref_image_1": [
    "101",
    0
   ],
   "ref_images.ref_image_2": [
    "102",
    0
   ],
   "ref_videos.ref_video_0": [
    "30",
    0
   ]
  }
 },
 "9": {
  "class_type": "BasicScheduler",
  "inputs": {
   "model": [
    "20",
    0
   ],
   "scheduler": "simple",
   "steps": 6,
   "denoise": 1.0
  }
 },
 "10": {
  "class_type": "MiniMaxH3TurboSampler",
  "inputs": {}
 },
 "11": {
  "class_type": "RandomNoise",
  "inputs": {
   "noise_seed": 760003
  }
 },
 "12": {
  "class_type": "BasicGuider",
  "inputs": {
   "model": [
    "20",
    0
   ],
   "conditioning": [
    "8",
    0
   ]
  }
 },
 "13": {
  "class_type": "SamplerCustomAdvanced",
  "inputs": {
   "noise": [
    "11",
    0
   ],
   "guider": [
    "12",
    0
   ],
   "sampler": [
    "10",
    0
   ],
   "sigmas": [
    "9",
    0
   ],
   "latent_image": [
    "8",
    1
   ]
  }
 },
 "14": {
  "class_type": "VAEDecode",
  "inputs": {
   "samples": [
    "13",
    0
   ],
   "vae": [
    "3",
    0
   ]
  }
 },
 "15": {
  "class_type": "VAEDecodeAudio",
  "inputs": {
   "samples": [
    "13",
    0
   ],
   "vae": [
    "4",
    0
   ]
  }
 },
 "16": {
  "class_type": "CreateVideo",
  "inputs": {
   "images": [
    "14",
    0
   ],
   "audio": [
    "15",
    0
   ],
   "fps": 24.0,
   "bit_depth": 8
  }
 },
 "17": {
  "class_type": "SaveVideo",
  "inputs": {
   "video": [
    "16",
    0
   ],
   "filename_prefix": "mmh3s3n4/MMH3S3N4_4_07_A",
   "format": "auto",
   "codec": "auto"
  }
 },
 "40": {
  "class_type": "ImageFromBatch",
  "inputs": {
   "image": [
    "14",
    0
   ],
   "batch_index": 51,
   "length": 56
  }
 },
 "41": {
  "class_type": "SaveImage",
  "inputs": {
   "images": [
    "40",
    0
   ],
   "filename_prefix": "mmh3s3n4_tail/4_07_A/f"
  }
 },
 "42": {
  "class_type": "ImageFromBatch",
  "inputs": {
   "image": [
    "14",
    0
   ],
   "batch_index": 106,
   "length": 1
  }
 },
 "43": {
  "class_type": "SaveImage",
  "inputs": {
   "images": [
    "42",
    0
   ],
   "filename_prefix": "mmh3s3n4_last_4_07_A"
  }
 },
 "100": {
  "class_type": "LoadImage",
  "inputs": {
   "image": "your_character_fullbody.png"
  }
 },
 "101": {
  "class_type": "LoadImage",
  "inputs": {
   "image": "your_character_face.png"
  }
 },
 "102": {
  "class_type": "LoadImage",
  "inputs": {
   "image": "your_secondary_character.png"
  }
 },
 "30": {
  "class_type": "VHS_LoadImagesPath",
  "inputs": {
   "directory": "E:\\sviluppo\\confyui\\outputs\\mmh3s3n4_tail\\4_06_A"
  }
 }
}