🎨 The AI Image & Video Creation Guide

🖼️ Event Posters & Invitations

Recommended Tools

Canva AI (fastest, has templates), ChatGPT (great for a complete poster concept with text baked in from a single description), or Adobe Firefly (best for photorealistic backgrounds you'll edit further).

Steps

1. Decide your one key message. A poster with five messages communicates none of them — pick the single most important thing (date, offer, or event name).
2. Generate or choose a background image using the prompt template below.
3. In Canva, add your text using a maximum of 2 fonts and 3 colours.
4. Check contrast — your text must be readable at thumbnail size, not just full screen.
5. Export at print resolution (300 DPI) if it's being printed, or web resolution if it's for social/digital.

📦 Product Images

Recommended Tools

Adobe Firefly for editing real product photos; ChatGPT, Flux, or Midjourney for fully AI-generated product scenes — ChatGPT is especially strong when you want accurate text, labels, or logos to appear correctly on the product.

Steps

1. Start with a clean, well-lit photo of your actual product if you have one — AI is much better at extending and enhancing a real photo than inventing a product from nothing.
2. Use Firefly's Generative Fill to replace or extend the background.
3. Generate 3–4 background variations before choosing — the first result is rarely the best.
4. Keep the product itself untouched; only regenerate the background and surrounding context.
5. Check for AI artefacts — warped edges, strange reflections, or text errors — before publishing.

📢 Ad Creative

Recommended Tools

Canva AI or ChatGPT for static ads; Runway, Kling, or Higgsfield for short video ads.

Steps

1. Write your ad copy first (see the Prompt Engineering Playbook, Asset Y02) — design should support the message, not the other way round.
2. Generate 2–3 visual directions and test which one supports your message most clearly.
3. For video ads, keep it under 15 seconds for social platforms — attention drops sharply after that.
4. Add captions to every video ad. Most people watch with sound off.
5. Create at least 2 versions (different hook, image, or angle) and compare performance once live.

🎬 Short-Form Video Content

Recommended Tools

Kling, Runway, Higgsfield, or Google Flow for generated footage; CapCut for editing, captions, and final assembly.

Steps

1. Write your script first using the SW1 or SW2 prompts in the Prompt Engineering Playbook.
2. Break the script into individual shots — generate each shot separately rather than asking for a whole video at once.
3. Generate 2–3 versions of your key shot before settling — the difference between usable and unusable is often just re-rolling the generation.
4. Assemble your clips in CapCut, add captions, music, and pacing.
5. Watch it back with sound off first — if it still makes sense, your visual storytelling is doing its job.

🤖 AI Influencer & Presenter Videos

Recommended Tools

HeyGen or Synthesia for the classic script-to-avatar workflow. Google Flow and Higgsfield are strong alternatives when you want your AI presenter placed inside a fuller generated scene rather than a plain studio background.

Steps

1. Write a short, conversational script (see SW3 in the Prompt Engineering Playbook) — sentences that sound natural spoken aloud.
2. Choose an avatar and voice that match your brand tone — energetic, calm, professional, playful.
3. Generate the video and check pacing — spoken AI avatars often need shorter sentences than you'd think.
4. Add your own branding, captions, and call-to-action text as an overlay.
5. ALWAYS disclose that the presenter is AI-generated, either verbally in the script or as an on-screen label. This isn't optional.

✅ Quality Checklist