How do I write good Runway prompts and settings?

Write prompts in English and describe one single continuous shot in the order "subject → action → scene → camera move → style". That structure beats piling up adjectives every time.

A prompt that works

` A young woman in a red coat walks through a snowy Tokyo street at night, neon signs reflecting on wet pavement, camera tracking shot from behind, cinematic, 35mm film look `

• Subject and action: a 5-10 second clip can hold one action. Avoid "first... then... finally..." multi-beat plots.
• Scene and light: time of day (at night / golden hour), weather, and environmental detail directly shape the mood.
• Camera language — Runway responds very well to it:

- static shot, slow zoom in / zoom out - tracking shot, pan left / right - drone shot / aerial view, handheld camera

• Style tags at the end: cinematic, 35mm film, anime style, hyperrealistic, slow motion.

Choosing the settings

• Duration: test compositions at 5 seconds first, then re-run the winning prompt at 10 seconds — it saves a lot of credits.
• Aspect ratio: 9:16 for vertical social clips, 16:9 for YouTube. The ratio can't be changed after generation; you'd have to re-run.
• Seed: fixing the seed in the settings keeps the overall look stable while you tweak the prompt, so you can iterate toward the shot you want.

Prompting for Image to Video

Once you've uploaded an image, the text only needs to describe motion and camera work — don't re-describe what's already in the picture. After uploading a portrait, she turns her head and smiles, gentle wind in her hair is enough.

Common problems

• Chaotic motion or warping: the action is too complex — cut it down to a single movement.
• Prompt seems ignored: non-English prompts work but are understood far less reliably; translating to English usually fixes it immediately.
• Faces falling apart: add close-up or switch to Gen-4, which handles faces much better than Gen-3.