🎬 Self-Learning Resource — Emenwa Global
Module 1: AI Prompting Fundamentals
Master the art of communicating with AI tools. Learn the universal prompt formula, when to use each text AI, and the special rules for every major video platform.
What Is an AI Prompt?
The foundation every creator must understand before touching any AI tool
An AI prompt is the specific instruction you give an AI model to produce a desired result. Think of it as a creative brief — the clearer and more specific your brief, the more accurately the AI delivers what you envision.
Prompts range from a single word to a detailed multi-paragraph document. The right length depends entirely on which AI tool you are using and what you are trying to create.
Key insight: Prompting is not the same across all tools. A text AI like ChatGPT reads prompts differently from an image AI like Midjourney, which reads them differently from a video AI like Veo 3 or Seedance. Each has its own logic — mastering each is what separates beginner creators from professionals.
Prompt Examples Across Tool Types
// TEXT AI (ChatGPT / Claude)
"Write a 3-minute motivational short film script about a
self-taught developer who overcomes self-doubt to launch
their first app. Cinematic tone. Emotional dialogue. Hopeful,
earned ending. Include camera direction suggestions."
// IMAGE AI (Midjourney)
"Portrait of a fictional female entrepreneur, late 30s,
confident expression, sharp blazer, modern glass office
behind her, golden hour light from the right, editorial
photography style, shallow depth of field —ar 16:9 —v 7"
// VIDEO AI (Veo 3)
"A teenage boy sits alone in an empty school library at dusk,
reading intently. A shaft of amber light falls across the book.
Camera: slow push-in toward his focused face over 8 seconds.
He looks up, half-smiles. Sound: quiet page-turn, distant
clock ticking, soft piano. Cinematic. 8 seconds, 16:9."
🔥 Golden rule: AI prompts always produce a result — but not always the same one. Even identical prompts can yield different outputs. Treat every prompt as a creative starting point. Experiment, iterate, and refine. That creative loop is the skill.
Text AI Tools for Scripts & Ideas
Which tool to use for which creative task — and how to prompt each one
ChatGPT — Versatile Creative Powerhouse
ChatGPT excels at narrative scripts, dialogue, brainstorming, and iterative refinement. Its biggest advantage is the conversational loop — you can revise outputs in real-time through back-and-forth conversation, refining until the result is exactly right.
// Script with Visual Cues
"Write a 2-minute video script for a cinematic short film.
A young mechanic in a small town discovers an old motorbike
hidden in a barn. Use emotional beats, minimal dialogue,
and strong visual storytelling. Include camera direction
notes and B-roll suggestions throughout."
// YouTube Series Outline
"Create a 6-part YouTube series called 'Quiet Inventors'
profiling self-taught creators who built something remarkable
with limited resources. Each episode: title, 3-minute story
arc, key interview questions, visual suggestions, hook line."
// Prompt Writing (using ChatGPT to write AI video prompts)
"Write a detailed Veo 3 text-to-video prompt for a scene
where an elderly clockmaker repairs the last mechanical clock
in a city that has gone fully digital. Include: lighting,
camera movement, sound design, and a single line of dialogue."
Claude — Emotional & Ethical Storyteller
Claude is best for scripts requiring emotional depth, psychological nuance, and ethically complex characters. It is the AI equivalent of a sensitive, thoughtful screenwriter — ideal for documentaries, social issue content, and character-driven drama.
// Social Issue Documentary
"Write a 5-minute short film script about a teenage girl who
discovers her small town's river is being poisoned by an
upstream factory and must decide whether to speak up despite
pressure from every adult around her. Show moral tension,
coming-of-age weight, and a realistic ending."
// Dual-Perspective Story
"Write a short film script told from two perspectives: an
AI system being decommissioned, and the engineer who built
it and has been ordered to shut it down. No villain. Just
two beings trying to understand each other in the last hour."
Gemini — Visual-First Planner
Google's Gemini thinks in terms of what the viewer sees — ideal for explainers, educational content, and product demos where text and visual elements need tight coordination.
// Visual-Structured Explainer
"Write a 90-second educational video script explaining how
the human immune system works for a general adult audience.
Include: voiceover narration, on-screen text suggestions,
visual cue descriptions for each section, camera angle
notes, and a memorable closing summary graphic."
Tool Comparison
| Tool | Best For | Strengths | Avoid Using For |
|---|---|---|---|
| ChatGPT | Versatile scriptwriting & dialogue | Fast, creative, iterative | Deep research accuracy |
| Claude | Emotional & ethical storytelling | Nuanced characters, long-form | Visual layout planning |
| Gemini | Visual-first explainers | Text + visual coordination | Deep emotional narrative |
| Perplexity | Research & fact-checking | Live citations & sources | Creative fiction writing |
| CoPilot | Technical coding tutorials | Step-by-step clarity | Drama or emotional arcs |
| Squibler | Professional screenwriting | Screenplay formatting | Quick idea generation |
| ChatSonic | Trending social content | Real-time web awareness | Long-form narrative |
The Universal Video Prompt Formula
The 8-part structure that works across every AI video tool
No matter which video AI you use — Veo 3, Kling AI, Runway, Sora, or Seedance — the best prompts follow the same core 8-part structure. Master this formula and every tool becomes easier to work with.
The 8-Part Formula:
[Shot Structure] → [Subject + Action] → [Setting + Time] → [Camera Movement] → [Style + Mood] → [Lighting + Colour] → [Audio + Dialogue] → [Negative Constraints]
[Shot Structure] → [Subject + Action] → [Setting + Time] → [Camera Movement] → [Style + Mood] → [Lighting + Colour] → [Audio + Dialogue] → [Negative Constraints]
// FULLY BUILT EXAMPLE — Using All 8 Parts
[SHOT STRUCTURE]
One continuous shot, 8 seconds, 16:9
[SUBJECT + ACTION]
A man in his 50s, wearing a weathered coat, stands at the
edge of a harbour wall at dawn. He holds an old photograph.
He closes his eyes and lets it go into the wind.
[SETTING + TIME]
Small fishing harbour. Early morning, blue hour fading into
first light. Mist on the water. Old wooden boats moored nearby.
[CAMERA MOVEMENT]
Start wide. Slow push-in over 8 seconds until we frame
his face in close-up as the photograph leaves his hand.
[STYLE + MOOD]
Cinematic drama. Melancholic but releasing. Quiet strength.
Photorealistic. Terrence Malick visual contemplation style.
[LIGHTING + COLOUR]
Blue-hour cool light warming slightly as dawn breaks.
Desaturated palette with warm amber in the horizon.
Soft natural lens flare as sun rises.
[AUDIO]
Distant harbour sounds — water lapping, a single gull.
Low cello note held beneath. No dialogue.
[NEGATIVE]
No cuts. No zoom. No watermark. No distorted face.
No artificial motion blur. Natural coat movement only.
Seedance Special Rules
Seedance (Higgsfield) uses a different priority system from other tools. The shot structure line must appear first, and the subject and core action must land within the opening 20–30 words. The model weights the beginning most heavily.
// SEEDANCE FORMAT — Shot structure first, subject immediately after
One continuous shot, 10 seconds, 9:16.
A surgeon in blue scrubs walks alone down an empty hospital
corridor at 3am, footsteps echoing. Overhead fluorescent lights
flicker in sequence ahead of her as she walks.
Setting: deserted modern hospital. Polished floors reflecting
the flickering light. A distant monitor beeps once.
Camera: locked tracking shot from 10 metres ahead, she walks
toward us. Slow, steady pace.
Clinical and eerie. Cold blue-white grade.
Negative: no cuts, no zoom, no camera wobble, no distorted face,
natural walking pace only.
VFX Bracket Notation for Seedance
When your scene includes special effects, embed them inline using [VFX: ...] notation rather than describing them separately. Seedance reads inline context more accurately.
// VFX inline — CORRECT for Seedance
She raises her hand and [VFX: shards of ice spiral outward
from her palm in slow motion, crystalline and blue-white,
catching the light as they fan across the frame].
// Separate block — LESS effective
[VFX section]
- Ice shards from palm
- Blue-white colour
(Seedance weighs narrative context, not separated lists)
Common Prompting Mistakes
- Vague descriptions: "A cool dramatic scene" → Be specific: "A chess grandmaster stares at the board for 90 seconds in total silence, then smiles and tips over his own king"
- Missing camera direction: Without it, AI defaults to static, uninspired framing. Always specify shot type and movement
- No negative prompts: Without explicit constraints, you will get watermarks, distorted faces, random cuts, and unwanted motion
- Ignoring audio: Veo 3 and Grok Aurora generate audio natively — always include sound design and dialogue instructions
- Over-packing one sentence: Break complex prompts into clearly labelled sections for maximum clarity
Module 1 — Knowledge Check
Test your understanding before moving on
1. When using Seedance, what must appear at the very beginning of your prompt?
A. The mood and colour grade description
B. The shot structure — count, duration, and aspect ratio
C. The negative constraints list
D. The VFX bracket notation
✅ Correct! Seedance weights the opening of the prompt most heavily. The shot structure line — count, duration, aspect ratio — must come first, immediately followed by the subject and core action in the opening 30 words.
❌ Not quite. The correct answer is B. Seedance reads the beginning of the prompt with the highest weight, so the shot structure (count, duration, aspect ratio) must appear first.
2. A director wants to write a script about a child coming to terms with the loss of a parent. Which AI text tool is the best choice?
A. ChatSonic — for trending real-time content
B. Perplexity — for research and citations
C. Claude — for emotional depth and nuance
D. CoPilot — for technical step-by-step clarity
✅ Correct! Claude is specifically built for emotionally intelligent, psychologically nuanced storytelling. It handles sensitive and deeply human themes with more care and depth than any other text AI tool.
❌ The best choice is C — Claude. It specialises in emotional depth, ethical complexity, and nuanced character writing, making it ideal for sensitive human stories.
3. Which part of the 8-part prompt formula prevents AI video tools from generating watermarks, distorted faces, and random cuts?
A. Style + Mood
B. Camera Movement
C. Shot Structure
D. Negative Constraints
✅ Correct! Negative constraints tell the AI what NOT to generate. Without them, tools default to their own choices — which often include watermarks, excessive cuts, motion blur, and distorted faces.
❌ The answer is D — Negative Constraints. This section explicitly tells the AI what to avoid, preventing common unwanted outputs like watermarks, distorted faces, and random transitions.