Text-to-Video Prompts: How to Write Prompts for AI Video
📞 +91 8075 400 500 · learn@cokonet.com Contact us
● AI filmmaking

Text-to-video prompts: how to write prompts that give you the shot you imagined.

Text-to-video prompts decide whether an AI video generator gives you a cinematic shot or a blurry mess. The model only knows what you write, so vague prompts waste credits and precise ones save hours. Good prompts read like a director's shot description: who, doing what, where, how the camera moves and how it is lit. This guide gives you a seven-part prompt formula, the camera and lighting words that matter, examples for common shots, and fixes for the problems every AI filmmaker runs into.

Cokonet Academy 12 November 20269 min read
Text-to-video prompts banner with a filmmaker typing a prompt while a screen shows an AI cinematic scene

How do you write a good text-to-video prompt?

Describe one shot in this order: subject, action, setting, camera (shot size and movement), lighting, style and mood, and any audio or duration. Keep it to one clear moment, use concrete nouns and verbs, and add one camera instruction. For example: "A young fisherman pulls a net onto a wooden boat at dawn on a Kerala backwater, medium shot, slow dolly-in, soft golden light and mist, cinematic, realistic."

Why prompts matter more in video than in images

Want to learn prompting across script, image and video tools? The AI Filmmaking course at Cokonet Academy covers prompt writing, consistent characters and editing in 70 hours.

Quick answer: Video adds motion and time. A prompt must describe what moves, how the camera moves and how the scene changes over a few seconds. Without that, models guess, and guesses produce jittery motion or random camera swings.

Each generation also costs more credits than an image. Spending two extra minutes on the prompt usually saves three or four failed generations.

The 7-part text-to-video prompt formula

Quick answer: Subject, action, setting, camera, lighting, style, and audio or extras. Not every prompt needs all seven, but checking each one stops most mistakes.

PartWhat to writeExample
1. SubjectWho or what, with key detailsA woman in a red cotton saree
2. ActionOne clear movementwalks slowly through a spice market
3. SettingPlace and timein old Mattancherry, late afternoon
4. CameraShot size and movementmedium tracking shot from the side
5. LightingSource and qualitywarm low sun through hanging fabric
6. Style and moodLook and feelingcinematic, realistic, calm
7. Audio or extrasSound, duration, aspect ratioambient market sounds, 8 seconds, 16:9

Camera and lighting words AI video models understand

Quick answer: Use standard film terms. Shot sizes: extreme wide, wide, medium, close-up, extreme close-up. Movements: static, pan, tilt, dolly-in, dolly-out, tracking, crane, handheld, drone shot, orbit. Lighting: golden hour, soft window light, hard noon sun, neon, backlit, rim light, low key.

  • Use one camera movement per shot; two often conflict.
  • Lens hints such as "35mm" or "shallow depth of field" shape the look.
  • Speed words help: "slow motion", "time-lapse", "gentle", "fast".
  • Mood words work best when paired with lighting: "tense, low-key lighting".

Example text-to-video prompts you can adapt

Quick answer: Start from a tested template and change the subject and setting. These examples cover product, travel, story and explainer shots.

  1. Product ad: "A glass bottle of coconut oil on a wet stone, water droplets sliding down, slow orbit shot, soft studio light with a warm rim light, clean commercial style, 6 seconds."
  2. Travel reel: "Drone shot gliding over Munnar tea gardens at sunrise, mist rolling between hills, smooth forward movement, cinematic, vivid greens."
  3. Short film moment: "Close-up of an old man's hands repairing a fishing net by lamplight, shallow depth of field, static camera, quiet and nostalgic, realistic."
  4. Explainer: "Animated 3D illustration of data flowing from a phone to cloud servers, clean blue and orange palette, slow camera pan, minimal style."
  5. Real estate: "Slow dolly-in through a bright modern living room in a Kochi apartment, afternoon light through tall windows, realistic, calm."

Image-to-video prompts and consistent characters

Quick answer: For image-to-video, do not describe what is already in the image. Describe only the motion, the camera and any change in light or mood. To keep a character consistent across shots, generate key frames from the same reference image first.

A practical pipeline is to design the character in an image model, create key frames for each shot, upscale them, then animate each with a short motion prompt such as "she turns her head toward the window, slow push-in, curtains move gently". Our comparison of the best AI video generators shows which models handle image-to-video best, and our guide on how to make an AI film covers the full process.

Fixing common AI video problems with prompts

Quick answer: Morphing faces, extra limbs, jittery motion and random camera swings usually come from too much action, conflicting instructions or vague subjects. Simplify the shot, specify one movement and use image-to-video for important characters.

  • Morphing or melting: reduce the amount of motion and shorten the clip.
  • Wrong camera: put the camera instruction early and use one movement.
  • Text looks garbled: add text later in the editor, not in the prompt.
  • Inconsistent style: reuse the same style words and a reference image every time.
  • Too generic: add specific objects, colours and textures.

You can see every creative programme on our all courses page or start at the Cokonet Academy homepage. Want feedback on your prompts in a live class? Book a free consultation or call +91 8075 400 500.

Prompt writing as a career skill

Quick answer: Agencies, brands and studios in Kerala now hire people who can turn a brief into consistent AI visuals and video. Prompt skill plus editing, story sense and client communication is what gets paid.

Keep a prompt library: every prompt that worked, the model, settings and result. It becomes your fastest tool and a portfolio asset. Our guide to starting a faceless YouTube channel with AI shows one way to put these skills to work.

Why Cokonet Academy for AI filmmaking

Cokonet Academy is an NSDC partner training institute with campuses in Trivandrum and Kochi. Find us on Google: Cokonet Academy, Trivandrum and Cokonet Academy, Kochi.

  • NSDC partner training institute with campuses in Trivandrum and Kochi, plus a Bengaluru centre and online-live batches
  • 11,200+ successful placements and 50+ corporate hiring partners
  • Expert trainers in SAP (FICO, MM, ABAP), AWS, Python with AI, Data Analytics and Power BI
  • Placement support until you are hired, with mock interviews and resume reviews

The AI Filmmaking course is built as one complete production pipeline: 70 hours across four modules, live and offline classes, 28 AI tools, portfolio projects and freelance pitching, taught by practitioners who make films with these tools.

Course details

  • Course: AI Filmmaking and Creative Production, Cokonet Academy
  • Duration: 70 hours across 4 modules: 10 hours fundamentals, 30 hours advanced AI tools, 10 hours video editing, 20 hours projects including freelancing
  • Mode: live and offline classes in Trivandrum and Kochi, plus online-live
  • Covers: storytelling and scripts with Gemini and GPT, prompt writing, shot lists and storyboards; consistent characters with Nano Banana, Seedream and Flux; text- and image-to-video with Seedance, Kling and Runway; upscaling with Magnific; voice and music with ElevenLabs and Suno; editing in VN Editor
  • Outcome: a portfolio of finished films and a freelance pitch, with a Cokonet completion certificate

Modules, tool stack and batch dates: AI Filmmaking course in Kerala.

Visit our training centres

Book a free consultation and sit in on a live session before you decide. We are open Monday to Saturday, 9:30am to 10:00pm.

Enrol now

Learn prompt writing, AI video generation and editing in one course. See the course and next batch, visit Cokonet Academy, or call and WhatsApp +91 8075 400 500.

Talk to a Cokonet counsellor - free, honest, and if we're not the fit, we'll say so.

Frequently asked questions

What is a text-to-video prompt?

A written description that tells an AI video generator what to create: subject, action, setting, camera, lighting and style.

How long should a text-to-video prompt be?

Usually one to three sentences describing a single shot; very long prompts often confuse the model.

What camera words work in AI video prompts?

Wide shot, medium shot, close-up, pan, tilt, dolly-in, tracking shot, orbit, drone shot and handheld.

How do I keep a character consistent in AI video?

Create the character in an image first and use image-to-video with the same reference image for every shot.

Why does my AI video look distorted?

Too much motion, conflicting instructions or vague subjects. Simplify to one action and one camera movement.

Should I describe the image in an image-to-video prompt?

No. Describe only the motion, camera and changes; the image already defines the look.

Can I add text inside AI-generated video?

It is unreliable; add titles and text in your video editor instead.

Do the same prompts work in every AI video generator?

The structure works everywhere, but each model responds differently, so test and adjust.

Can I use AI to write video prompts?

Yes. Ask ChatGPT or Gemini to expand your idea into the 7-part formula, then edit it yourself.

Where can I learn AI video prompting in Kerala?

Cokonet Academy's 70-hour AI Filmmaking course in Trivandrum, Kochi and online-live teaches prompting, AI video tools and editing.

Keep reading