{
 "slug": "guide-seedance-2-5",
 "name": "Prompt guide: Seedance 2.5",
 "category": "guide",
 "group": "Prompt guides",
 "media": [
  "video"
 ],
 "checked": "2026-10-13",
 "effect": "Scenes of up to 30 seconds that hold their characters, look and timing.",
 "use_when": "Multi-shot scenes, dialogue, ads with a story, edits on real footage.",
 "phrase": "one continuous block in labelled sections: the look first, the scene, the cast, the place, the first frame and who stands where, then each shot on its own timecode, then optics, camera, physics, light, and the sound last",
 "pitfalls": [
  {
   "text": "Skip a section and the output fails in a predictable way; the tested prompts include all of them.",
   "evidence": "vendor",
   "source": "Higgsfield Seedance 2.5 prompting guide, read 28 Sep 2026"
  },
  {
   "text": "When the lines come back scrambled or the speech speeds up, the prompt asks for more than the duration holds; split the scene.",
   "evidence": "practitioner",
   "source": "Rourke Heath (25 Sep 2026) via Director Chair field notes on video prompting"
  }
 ],
 "per_model": [],
 "external_refs": [],
 "licensed": [],
 "site_names": [],
 "related": [
  "which-model-for-video",
  "global-look-block",
  "cast-block",
  "screen-direction"
 ],
 "rights": "vendor-summary",
 "evidence": "vendor",
 "source": {
  "url": "https://higgsfield.ai/blog/seedance-2-5-prompting-guide",
  "note": "Own words from Director Chair field notes on video prompting and the Higgsfield guide summary"
 },
 "body": "## The prompt, in order\n\n1. **Global look:** camera, lens, stock, light, grade, one block, the same words in every prompt of the job.\n2. **Scene:** what happens, in two sentences.\n3. **Cast:** one line per person with only what never changes, plus their voice.\n4. **Place:** the location, its light, what is in it.\n5. **First frame and blocking:** who stands where, as positions in the frame.\n6. **Shots:** each with its own timecode, framing, action and the kind of cut into the next.\n7. **Optics and camera:** field of view in degrees plus distance; one main move per shot.\n8. **Physics and light:** what changes one way only, what repeats as a rule, where the light comes from.\n9. **Sound:** ambience, effects, music, the spoken lines with their delivery.\n\nEvery rule that must never break gets its own named lock (how many people exist, who stays on which side).\n\n## Lengths\nOne generation holds about 30 seconds. A longer script is split at a hand-off into parts that cut together, each carrying the same look and cast blocks.",
 "shown_with": null,
 "url": "https://directorchair.ai/guide/guide-seedance-2-5",
 "markdown": "https://directorchair.ai/guide/guide-seedance-2-5.md",
 "prompts": {
  "seedance-2.5": "Scene: [your subject] in [your setting].\nShot 1 (0:00 to 0:05): one continuous shot",
  "seedance-2.0": "Scene: [your subject] in [your setting].\nShot 1: one continuous shot"
 },
 "licence": {
  "code": "MIT",
  "text": "CC BY 4.0",
  "third_party": "Clips and card texts inside a \"licensed\" block belong to their makers and are shown with permission and credit."
 }
}