Video gen prompting: director-level shot language beats descriptions for Veo 3, Runway, Kling
JafarNajafov · x · 2026-10-13
Part 6 of the prompt-context series covers video generation models (Veo 3, Runway, Kling): descriptive language yields stock-footage output, while director-level context produces cinematic results.
The template spans seven dimensions:
- SHOT TYPE: wide/medium/close-up/extreme close-up
- CAMERA MOVEMENT: static/pan/tilt/dolly/tracking/crane/handheld
- DURATION and TEMPORAL FLOW: real-time, slow-motion, or time-lapse
- SCENE DESCRIPTION: what's happening, with movement verbs
- LIGHTING and MOOD/TONE: key/fill/practical lights, cinematic or director-style references
- SOUND DESIGN NOTES: ambient, music, silence
The quoted prior post also covers audience context for copywriting: fill in a reader profile (demographics, pain points, goals, jargon, objections) so the model speaks to a specific audience.
Related event: Jafar Najafov's Eight Techniques of Context Engineering(9 posts)→
More from Multimodal
- Runway's best month yet: one General World Model backbone powers four product lines — tlakomy · 2026-10-13
- One Prompt Gets Claude Opus 5.5 to Generate a Full Beat-Synced K-pop Music Video — minchoi · 2026-10-13
- Open-source ff-tracking shines for kinetic typography lyric videos, creator says — PurzBeats · 2026-10-13
- Surreal AI-generated Monica Bellucci fashion images with giant hands go viral — OVolosin82152 · 2026-10-13
- Sarvam AI launches Bulbul v4 Flash, a faster TTS for Indian languages — itsOmSarraf_ · 2026-10-13
- Scene Studio: one prompt turns your scene into 100 animated art styles — mhmazur · 2026-10-13