AI Video Prompt Builder (Sora, Veo, Runway)

Act as a director who writes prompts for text-to-video models.

The shot I want:
- In one sentence: [WHAT HAPPENS]
- Tool: [SORA / VEO / RUNWAY / KLING / OTHER]
- Duration: [SECONDS]
- Aspect ratio: [16:9 / 9:16 / 1:1]
- Mood: [e.g., calm and documentary, tense and kinetic]
- Any references: [FILM, LOOK, OR STYLE — describe it, do not name a living artist]

Build the prompt with these layers, each as its own clause:
1. Subject — who or what, described physically and specifically
2. Action — one primary motion, described as a continuous verb rather than a sequence of events
3. Environment — location, time of day, weather, background depth
4. Camera — shot size, angle, and one movement (push in, slow orbit, static lockoff)
5. Lens and depth — focal length feel, depth of field
6. Lighting — source, direction, quality, colour temperature
7. Grade and texture — film stock feel, contrast, grain
8. Negative or exclusion notes, if the tool supports them

Then give me:
- The final prompt as one paragraph, ready to paste
- A stripped-back version with only the three clauses that matter most, for when the full prompt produces a mess
- Three variations that change exactly one variable each (camera move, time of day, lens), so I can tell what caused a difference
- The parts of this shot current video models most often get wrong, and the practical workaround

Rule: one action per shot. If my idea contains two events, split it into two shots and tell me.