Phase 6: Image, Video & Voice Prompting · 45 min · Midjourney/DALL-E, Sora/Runway (optional), ElevenLabs (optional)

Project 3: Visual Campaign

A consistent set of AI images or a short video following a style guide you wrote. Show the prompts, the iterations, and the final output.

Real-world application: What changes when you use this: you have a visual campaign produced with AI — proof you can create consistent, professional visual assets, which is a marketer's dream skill.

What you will learn

Produce a consistent set of AI images or a short video following a self-written style guide
Document the prompt iterations from first attempt to final output
Apply image/video prompt anatomy to a real project
Create campaign-ready visual assets

Objective

What you're building

A visual campaign — a set of 4-6 consistent AI-generated images OR a short (10-15 second) AI-generated video — that follows a style guide you wrote and tested yourself.

This is the third artifact in your prompt portfolio.

Choose your format

Option A: Image Campaign

Create 4-6 images that look like they belong together. Examples:

A set of product photos for a fictional brand
A series of character portraits in a consistent art style
A set of social media graphics with a consistent visual identity
A mini lookbook for a fashion concept

Option B: Video Campaign

Create a short (10-15 second) AI-generated video with:

A clear subject, action, and environment
Intentional camera movement and lighting
(Optional) Voiceover or music

The process

Step 1: Define Your Campaign

What is it for? (A fictional brand, a social media series, a concept project)
What's the visual identity? (Color palette, mood, style, composition approach)
What are the 4-6 subjects/scenes?

Step 2: Write Your Style Guide

Write a style anchor (reusable style description)
Write a negative prompt for common issues
Note any tool-specific settings (aspect ratio, quality, version)

Step 3: Generate and Iterate

For each image/scene:

Write the prompt (subject + composition + lighting + style + mood + style anchor)
Generate
Assess: does it match the style guide? Is the subject right?
Iterate: adjust the prompt and regenerate
Document the iterations

Step 4: Review for Consistency

Look at all 4-6 images together:

Do they look like they belong together?
Is the style consistent?
Is the quality consistent?
If not, what needs to change? Adjust and regenerate.

Step 5: Document Everything

Deliverable format

# Visual Campaign
[Your Name] | [Date]

### Campaign Brief
- Purpose: [what it's for]
- Visual identity: [color palette, mood, style]
- Subjects: [list of 4-6 images/scenes]

### Style Guide
### Style Anchor
[Your reusable style description]
### Negative Prompt
[Your negative prompt]
### Settings
[Tool-specific settings]

### Image 1: [Subject]
### Final Prompt
[The prompt that produced the final image]
### Iterations
1. [First attempt prompt] → [What was wrong]
2. [Second attempt prompt] → [What was wrong]
3. [Final prompt] → [Why this worked]
### Final Image
[Image or description]

### Image 2: [Subject]
...

(continue for all images)

### Consistency Assessment
[How consistent are the images? What could be improved?]

### What I Learned
[2-3 sentences on what this project taught you about visual prompting]

Why this matters

This project proves you can:

Write and apply a consistent visual style guide
Use concrete visual vocabulary to control image output
Iterate on prompts to get from "close" to "right"
Produce campaign-ready visual assets

In Phase 9, this becomes part of your portfolio. Even now, it's proof that you can create professional visual assets with AI — a skill that's increasingly valuable in marketing, content creation, and design.

Your practice rep (project)

Complete the full visual campaign. Save the document. This is your third portfolio artifact.

Build

Do Project 3: Visual Campaign (the steps are under Your practice rep below) as a before-and-after test. Keep the prompt you would normally write and the answer it gets. Change one thing — the technique from this lesson — and run the same task again. Before you read the new answer, write down what a good answer must do, then judge both against that list. Note at least one way the new version did worse or still fell short; that is as useful as the improvement.

Save your work as visual_campaign.md and prompt-experiment.md: the lab asks for them, and you check them against its rubric.

Unlock the full lesson

You've read the first 2 sections. The rest of this lesson covers Check Yourself, Key Terms & Next — plus a hands-on lab, quiz, and project artifact.

Create a free account — the opening phases of 29 of 35 courses are free, no credit card.