HALLOWEEN AI VIDEO PROMPTS FOR REELS AND SHORTS - README
Source: https://gauravtiwari.org/halloween-ai-video-prompts/

WHAT IS IN THIS ZIP
- README.txt: this file, with how to use the prompts and a blank template
- 1-candlelit-portrait.txt: the candlelit portrait scene
- 2-sheet-ghost.txt: the sheet ghost scene
- 3-pumpkin-lantern.txt: the pumpkin lantern scene
- 4-foggy-witch.txt: the foggy witch scene
- all-prompts.txt: all 4 scenes in one file

HOW TO USE IT
Make or pick the starting image first, then upload it to an image-to-video tool as the
first frame. Paste the MOTION and CAMERA lines into the prompt box. If the tool has its
own camera menu, set the move there and leave the CAMERA line out. Add the REPAIR line
at the end of the prompt. If the tool has a separate negative prompt field, put only the
unwanted things there as a plain list, for example: extra people, text, scene change.
Leave out the word "no" in that field.

SETTINGS FOR EVERY SCENE
- Aspect ratio: 9:16 (vertical)
- Length: the shortest option the tool offers for a first try
- Audio: off, or generate it and replace it later with licensed music in Instagram or YouTube
- Change one line at a time when you regenerate

FIELD TEMPLATE (copy this for your own Halloween scene)

SCENE:
Vertical 9:16 photograph of [one subject] in [one simple setting], [light source].
[Framing]. No text, no logos.

MOTION:
[Subject] [one slow action]. [Light or weather that moves]. [What stays still].

CAMERA:
Static camera / slow push-in / slow pull-out / slow pan [left or right].

REPAIR:
Keep [face, shape or object count] exactly as in the first frame. No new people, no
text, no cuts.

WHAT EACH FIELD CONTROLS
- SCENE: the subject, setting, light and framing of the starting image.
- MOTION: what moves, how much, and what stays still.
- CAMERA: how the frame itself moves.
- REPAIR: what must not change from frame to frame.
