video-storytelling-core-principles

Check each scene against the mute test: mute the narration and verify the setup, turn, and payoff still read through action contrast alone. Then confirm failures and successes have visible physical causes rather than abstract labels.

The mute test requires that the picture tells the story without audio. Voiceover should annotate the image, not carry it; problem and solution must be shown through visual contrast, such as a character struggling solo versus working calmly with help.

Keep Mandarin voiceover at or below roughly 4.5 characters per second. Cut copy before cramming frames so there is time for physical acting, sound effects, and the viewer's eye to rest.

Avoid arbitrary black frames or hard cuts. Bridge shots with moving or glowing elements already in frame, such as a flour speck whiting out the lens into a new scene or a glowing icon expanding to reveal the next shot.

Require stable identity in diffusion video prompts: same character, outfit, hairstyle, and framing unless a deliberate occlusion or scene reset occurs. Add an explicit style line such as "same person, same clothes, same hairstyle as previous shot."

Abstract buzzwords like "system" or "empowerment" cannot be visualized directly and produce vague staging. Map ideas to concrete everyday scenes with physical texture, such as kitchen sprites dividing cake work to represent multi-agent teamwork.