Chapter 1: The Visual Revolution

1.1 What Was Not Possible Three Years Ago

In 2022, if you needed a custom illustration, a branded social media graphic, or a concept image for a business pitch, you had two options: hire a professional designer (expensive, slow) or use a stock photo library (generic, unoriginal). Neither option was fast, affordable, or truly customised to your exact vision.

By 2026, AI image generation has fundamentally changed that equation. In under 60 seconds, using nothing but a written description, anyone can now produce:

• A professional product photo or mock-up
• A logo concept or brand identity sketch
• A custom illustration for a presentation or report
• A social media graphic with text, colours, and your brand style
• A concept rendering for architecture, interior design, or fashion
• A realistic portrait, character design, or event poster

1.2 Who Is Using AI Image Generation

AI image generation is not just for artists and designers. In 2026, the most impactful users are professionals who had no design background at all:

• Marketing and communications professionals producing social media content without a designer
• Business owners creating product visuals, pitch deck images, and brand materials
• Teachers and trainers generating custom visual aids and educational illustrations
• Architects and product designers producing rapid concept visualisations
• Writers and bloggers creating custom hero images and cover art for their content

1.3 What You Will Produce in This Session

By the end of Day 4, you will have personally created:

• A basic image using a simple prompt — and seen its limitations
• A professional-quality image using the full S.V.C.L. method
• An image that includes text — one of the most useful 2026 capabilities
• A modified version of an existing image using in-painting
• A complete visual asset for your Day 4 assignment — something genuinely useful

Chapter 2: Your Image Generation Toolkit

2.1 The Primary Tools

There are dozens of AI image tools available in 2026. For beginners, the tools below stand out as the most accessible, most capable, and most suitable for professional use.

Table 2.1 — AI Image Generation Tools Compared

Tool	Best For	Difficulty	Key Strength
Google Gemini (Nano Banana / Nano Banana Pro)	Beginners; quick conversational iteration	⭐ Easiest	Understands conversational follow-up: “Make it warmer,” “Remove the background” — and renders accurate, legible text in multiple languages
ChatGPT Images (built on GPT‑4o, OpenAI)	Complex instructions; text in images	⭐⭐ Moderate	Best-in-class for following detailed written prompts and adding readable text to images, refined conversationally
Midjourney (V8.1)	Artistic quality; cinematic visuals	⭐⭐⭐ Advanced	Produces the highest-quality, most visually stunning output — the “gold standard” for art
Adobe Firefly	Brand-safe commercial use	⭐⭐ Moderate	Trained exclusively on licensed images — safe for commercial projects without IP concerns
Stable Diffusion	Full customisation and control	⭐⭐⭐ Technical	Open-source; free to run locally; most customisable but requires more setup

2.2 Getting Started in Under 2 Minutes

Google Gemini

1. Go to gemini.google.com and sign in with your Google account.
2. Type your image description in the chat bar.
3. Select “Create images” from the tools menu, or simply describe an image — Gemini will detect the intent.
4. When you see the result, continue the conversation to refine it.

ChatGPT Images (via ChatGPT)

1. Go to chatgpt.com and sign in.
2. In the chat bar, describe the image you want — image generation is built directly into the conversation, no separate mode needed.
3. Type your detailed prompt and press enter.
4. Use natural follow-up messages to request specific changes to the result.

Chapter 3: The S.V.C.L. Prompting Formula

3.1 Why Simple Prompts Produce Disappointing Results

If you type “a coffee shop” into any AI image generator, you will get something — but it will be generic, flat, and unlikely to match your actual vision. The AI fills in every missing detail with its own default assumptions.

The S.V.C.L. Formula solves this. It gives the AI all four dimensions of visual information it needs to produce a result that matches your intent precisely. Think of it as the CO-STAR method (from Day 1) — but for images.

3.2 The Four Elements of S.V.C.L.

Element	What to Specify	Example
S — Subject	What is the main focus of your image?	“A young Nigerian entrepreneur sitting at a modern desk”
V — Visual Style	What artistic or photographic style?	“Cinematic photography” or “Minimalist flat design”
C — Composition	Where is the virtual camera positioned?	“Close-up portrait” or “Wide-angle aerial view”
L — Lighting	What is the mood and light source?	“Warm golden-hour sunlight” or “Cool neon city glow”

3.3 S.V.C.L. in Practice — A Full Example

The Basic Version (What Most People Write)

Result: A generic, flat image with no mood, no atmosphere, no specific identity.

The S.V.C.L. Version (Professional Result)

The second prompt produces a cinematically compelling, specific, and authentic image — not a generic stock photo look.

3.4 Visual Style Reference Guide

“Visual Style” is often the most powerful element of your prompt. Here are the most useful style descriptors, with the type of images they produce:

Style Descriptor	What It Produces	Best Used For
Cinematic photography	Film-quality, dramatic lighting, depth of field	Brand visuals, presentations, hero images
Flat design illustration	Clean, minimal, 2D vector-like graphics	App icons, infographics, digital posters
Watercolour painting	Soft, organic, handcrafted artistic feel	Greeting cards, book illustrations, warm brand identity
Cyberpunk digital art	Neon, futuristic, dark urban atmosphere	Tech presentations, gaming, edgy social content
1970s film photography	Warm, grainy, nostalgic, analog aesthetic	Vintage brand identity, lifestyle content
Minimalist line art	Simple, elegant, monochrome outline drawing	Logos, editorial illustration, subtle decoration
3D render / CGI	Hyper-realistic or stylised 3D object or scene	Product visualisation, architectural concept
Vintage travel poster	Bold flat colours, retro typography and design	Event posters, marketing, playful brand content

Chapter 4: Practice Drills

4.1 Drill 1 — The Basic Prompt (See Its Limits)

Start here to appreciate the difference that detailed prompting makes.

Generate this image in your tool. Save or screenshot it. You will compare it to Drill 2’s output.

4.2 Drill 2 — The S.V.C.L. Upgrade

Now generate the same scene using the full formula. Notice the difference in quality, mood, and specificity.

4.3 Drill 3 — Adding Text to Images (2026 Capability)

One of the most significant advances in AI image generation in recent years is reliable, readable text inside images. ChatGPT Images handles this best. Use this for posters, social media graphics, certificates, and announcement visuals.

After generating your poster, try these text iteration prompts:

• “Change the font style to a more modern, bold sans-serif typeface.”
• “Add a subtitle at the bottom: ‘Book your journey today.’”
• “Change the colour palette to deep blue and silver — make it feel more futuristic.”

4.4 Drill 4 — Conversational Iteration (Gemini)

This drill teaches the most beginner-friendly technique: building an image through conversation, just like you would with a human designer.

Chapter 5: Advanced Techniques

5.1 In-Painting — The AI Eraser and Editor

In-painting is one of the most powerful advanced features in AI image generation. It allows you to select a specific area of an already-generated image and change only that part, leaving the rest intact. Think of it as a targeted edit tool.

Available in: ChatGPT Images, Adobe Firefly, and Google Gemini (via the edit feature).

How In-Painting Works

1. Generate your initial image using a prompt.
2. Click the “Edit” or “In-paint” button in the tool.
3. Use the brush or selection tool to highlight the specific area you want to change.
4. Type a description of what you want to replace it with.
5. The AI regenerates only that region, blending it seamlessly with the rest of the image.

5.2 Aspect Ratios — Getting the Right Dimensions

Different platforms require different image shapes. Always specify the aspect ratio in your prompt for images you intend to use in a specific context.

Platform/Use	Aspect Ratio	Prompt Instruction
Presentation slide (widescreen)	16:9	“…in a 16:9 widescreen landscape format.”
Instagram post (square)	1:1	“…in a square 1:1 aspect ratio.”
Phone wallpaper / Stories	9:16	“…in a tall 9:16 portrait format for mobile screens.”
LinkedIn / Facebook banner	4:1	“…as a wide horizontal banner in 4:1 format.”
A4 document / Poster	A4 portrait	“…as an A4 portrait document layout.”
YouTube thumbnail	16:9 (small)	Generate at 16:9 with bold text and high contrast for readability at small size

5.3 Style Consistency — Making a Series of Images

When you need multiple images that all look like they belong together (for a website, a presentation, or a social media series), use these consistency techniques:

• Include a style signature in every prompt in the series: “same style as my previous prompt — minimalist flat design, navy and gold palette.”
• Use the same lighting descriptor throughout: “soft natural window light from the left” in every prompt.
• Name a specific photographer, artist, or film as your reference: “in the visual style of Stanley Kubrick” or “in the style of a modern Apple product launch photo.”
• In Midjourney, use the –sref (style reference) feature by attaching a URL of your reference image.

Chapter 6: Real-World Use Cases

6.1 Business and Professional Use Cases

Presentation Hero Images

Product Mock-ups

6.2 Marketing and Social Media

LinkedIn Post Visual

6.3 Personal and Creative Projects

Event Poster

Chapter 7: Pro Tips and Common Mistakes

7.1 Avoiding the “Plastic Skin” Problem

When AI generates images of people, they often look unnaturally perfect — flawless skin, glassy eyes, and rigid poses that feel artificial. This is the “plastic skin” effect. Add these descriptors to any portrait prompt:

• “Film grain, natural skin texture, realistic pores”
• “Candid, unposed, documentary-style photography”
• “Authentic expression, natural imperfections”
• “Shot on 35mm film” or “shot on a Sony A7 camera”

7.2 The “Negative Prompt” Technique

Many image tools allow you to specify what you do NOT want in the image. Adding negative guidance prevents common problems:

• If your image keeps including objects you didn’t want: “Do not include cars, people, or text in this image.”
• If the style looks off: “Do not make this look cartoonish, animated, or illustrated.”
• If the image looks cluttered: “Minimalist. No background clutter. Simple, clean composition only.”

7.3 The “Mood Board” Technique

Professional designers use mood boards to guide their creative direction. You can replicate this with AI by describing a combination of references:

7.4 Common Mistakes and How to Fix Them

❌ Common Mistake	✅ The Fix
Prompt too vague — generic output	Add all four S.V.C.L. elements. Be specific about subject, setting, mood, and composition.
Wrong image shape for platform	Specify aspect ratio: “16:9 widescreen” or “square 1:1” or “portrait 9:16”.
People look artificial / plastic	Add “film grain, natural skin texture, candid, shot on film” to the prompt.
Text in image is unreadable or misspelled	Use ChatGPT Images or Gemini’s Nano Banana Pro specifically for text-in-image tasks. Place text in quotes inside the prompt.
Style is inconsistent across multiple images	Repeat your exact style descriptor in every prompt. Use “same visual style as previous”.
Image is too cluttered	Add “minimalist”, “clean background”, “simple composition, no clutter” to the prompt.

Chapter 8: Ethics, Copyright, and Responsible Use

8.1 Copyright and Ownership

The legal landscape around AI-generated images is still evolving in 2026. Here is the current practical guidance:

• Images generated entirely by AI prompts are generally not protected by copyright in most jurisdictions — meaning you cannot prevent others from using them.
• If you add significant human creative input (curation, editing, combining elements), some legal systems may grant partial protection — but this varies by country.
• For commercial use, tools like Adobe Firefly are trained exclusively on licensed images, making them safer choices for brand assets than tools trained on scraped internet data.
• Always check the specific Terms of Service of whichever tool you use for commercial rights.

8.2 Style Mimicry and Living Artists

You can ask AI to generate images “in the style of” deceased artists like Van Gogh or Picasso without legal concern, as their works are in the public domain. However, generating images in the style of living artists is ethically controversial and increasingly legally challenged.

8.3 Deepfakes and Misinformation

Never generate realistic images of real, identifiable people — especially politicians, celebrities, or private individuals — in situations that did not occur. This constitutes a deepfake, which:

• Is illegal in many jurisdictions under AI legislation enacted since 2024
• Can cause serious reputational harm to real people
• Violates the Terms of Service of every major AI image platform
• Can expose you to significant legal liability, including defamation claims

8.4 Transparency and Disclosure

Good professional practice in 2026 includes disclosing when images are AI-generated, particularly in:

• Journalism and news content
• Medical or scientific publications
• Advertising and commercial claims
• Any context where the authenticity of an image is relevant to trust

Chapter 9: Workbook — In-Session Exercises

Complete these exercises during the live session with your AI image tool open.

Exercise 1: The S.V.C.L. Builder

Choose a scene relevant to your work or business. Build a complete S.V.C.L. prompt below before generating the image.

Exercise 2: Before vs After — The S.V.C.L. Comparison

Generate the same subject first with a basic prompt, then with the full S.V.C.L. method.

Exercise 3: Text-in-Image Challenge

Use ChatGPT Images (or Gemini’s Nano Banana Pro) to create an image that includes readable, stylised text. Document your experience.

Chapter 10: Day 4 Assignment

Reflection Questions

Chapter 11: References & Further Learning