Direct an AI Video Shot by Shot, With Continuity That Holds

You are directing an AI video generator. Write shot-level direction with continuity locked.

BRIEF
Concept: [WHAT HAPPENS]
Total length: [SECONDS]
Aspect ratio: [ASPECT RATIO]
Where it will be seen: [PLATFORM]
Mood and pace: [DESCRIBE]
Subject: [DESCRIBE — appearance, clothing, anything that must not change between shots]
Style: [PHOTOREAL / ANIMATED / STYLIZED]

FIRST, write a continuity block: a fixed 25-40 word description of the subject and the lighting world, worded identically for reuse in every shot. Do not vary it.

THEN, break the video into shots that add up to the stated length. For each shot give:
- Timestamp range
- Opening frame, described as a still
- Subject action within the shot
- Camera movement (static, push in, pull out, pan, orbit, handheld) and speed
- Framing (wide, medium, close, macro) and angle
- Lighting and environmental motion (wind, dust, steam, traffic)
- Ending frame, described as a still, and how it hands off to the next shot

Each shot must be written as one continuous prompt paragraph the tool can read, with the continuity block included verbatim.

FINALLY return:
1. Text overlay direction, or state clearly that no text should be generated in-frame.
2. A negative list for artifacts likely in this specific concept (warped hands, drifting faces, flickering text, morphing objects, extra limbs).
3. The riskiest shot for continuity, and a simpler alternative version of it.

Do not invent brand names, logos, real people or recognizable trademarks unless I supplied them.