minimax/h3-max/reference-to-video

{
  "video": {
    "url": "https://v3b.fal.media/files/b/0aa91f5a/mMxt2Ga15hBB51PNsOdYo_minimax-h3.mp4",
    "content_type": "video/mp4",
    "file_name": "mMxt2Ga15hBB51PNsOdYo_minimax-h3.mp4",
    "file_size": 5252129
  },
  "expanded_prompt": "subject_definitions:\n<Subject 1> is the young man with curly dark hair wearing an olive green jacket and a white t-shirt, as seen in <Picture 1>.\n<Subject 2> is the middle-aged man with salt-and-pepper hair and a mustache, wearing a tan canvas jacket over a blue shirt, as seen in <Picture 2>.\n\nsummary:\n[reference generation] The target video shows a conversation between <Subject 1> and <Subject 2> standing in an open asphalt lot, featuring synchronized dialogue and subtle emotional reactions.\n\nretention_analysis:\n<Subject 1> (appears in [Shot 1]): fully_preserved - maintains facial features, curly hair, and olive green jacket from <Picture 1>.\n<Subject 2> (appears in [Shot 1]): fully_preserved - maintains facial features, mustache, and tan canvas jacket from <Picture 2>.\n\ndetailed_description:\nThe video is rendered in a cinematic style with a naturalistic, slightly desaturated color palette, mirroring the lighting and atmosphere of the reference images. The scene takes place in a wide, empty asphalt parking lot under soft, overcast daylight.\n\n[Shot 1] The scene is a medium over-the-shoulder shot, focusing on <Subject 1> in the foreground (near) and <Subject 2> in the background (far). The camera is a Static Shot with no movement. <Subject 1> (S1) looks toward <Subject 2> and says flatly, <d>[English] Did you hear about H3 Max?</d>. After the line, there is a brief beat of silence where neither man speaks, and the only movement is the slight rustle of their clothing. <Subject 2> (S2) maintains a neutral expression for a moment before his mouth pulls into a slow, subtle half-smile. <Subject 2> (S2) then answers in a warm and certain tone, <d>[English] Dude, I love it.</d>. Upon hearing this, <Subject 1> (S1) lets out a soft exhale and shakes his head once in a slow, slight motion, his expression softening as if almost smiling. The lip-sync is precise and natural for both speakers.\n\noverall_soundscape:\nThe audio environment is an open outdoor setting. It features a constant, low-frequency distant traffic hum and the intermittent whistling of wind moving across the empty lot.\n\nnon_diegetic_music:\nN/A",
  "seed": 1851572118,
  "timings": {
    "inference": 4.423960474989144
  }
}