🎨 Image generation — the real guide
The gap between a mediocre AI image and a genuinely useful one is almost entirely in the prompt. Most people type a sentence. The good ones write a brief.
Here's what the tools look like right now:
The anatomy of a good image prompt
A good image prompt has four components. You don't need all four every time, but knowing them changes your results dramatically.
The rule: Be the art director, not the requester. Instead of "a photo of a product," write the brief you'd give a photographer. Lighting, angle, mood, context, camera feel.
🎵 AI music — Suno and Udio properly
Suno (suno.ai) generates complete songs — lyrics, vocals, instrumentation, production — from a text description. Udio (udio.com) does the same with slightly different strengths (often better at genre authenticity). Both have free tiers.
The output quality has crossed a threshold: the results are genuinely listenable, often impressive, and occasionally astonishing for what a text prompt produces. Here's how to get results worth using.
Suno — what works
For the lyrics field, Suno takes either free text or structured lyrics. If you want control over what gets said, write the lyrics yourself using Claude first:
Copyright note: Suno and Udio own the commercial rights to outputs on free tiers. Paid tiers give you commercial rights. If you're using it for a business — advertising, background music for a venue, products — get clear on the licence before you publish.
🎬 Video generation — what's actually good right now
Video generation is moving faster than any other AI creative tool category. What was impressive six months ago is already the baseline. Here's the honest state of play:
Honest assessment: Video AI is still imperfect. Hands, faces, physics, and text all break. The best use cases right now are: abstract/stylised content, product B-roll, animating illustrations, and generating texture/background loops. It's not ready to replace a camera operator — but it's genuinely useful for things that would otherwise not get made.
For a Raglan-based creative: think about animating a gig poster into a short social video, turning a flat product photo into a moving loop, or generating abstract ocean textures for a title sequence.
✍️ AI for writing — beyond basic drafts
Module 2 covered the basics. This goes further: specific use cases for artists, musicians, and small businesses in Raglan.
Grant applications — probably the highest-value use of AI writing for the Raglan arts community. Arts Foundation NZ, Creative NZ, Waikato Regional Council all have funding rounds. AI can help with structure, tone, and not starting from a blank page.
Artist statements — the hardest writing most artists ever do. AI is genuinely useful here for getting a first draft you can then make your own.
The rule for AI-assisted writing: Use it to start and structure, then rewrite in your own voice. The goal is to not stare at a blank page — not to avoid writing. Your voice is the product, not the AI's output.
🤔 Your creative voice in an AI world
The anxiety is real: if AI can generate a plausible version of your style, what is your work worth? It's worth engaging with this directly rather than either dismissing it or catastrophising.
What AI does well: generating competent, average versions of existing styles. Give it "impressionist sunset" and it produces something that looks like an impressionist sunset. Give it "folk song about loss" and it produces something that sounds like a folk song about loss.
What AI cannot do: have a life. Your work is shaped by where you grew up, what you've survived, who you've loved, what you've noticed. Raglan at 6am in winter is not a prompt result — it's yours. AI can approximate the visual grammar of that image; it cannot have been there.
The artists who are thriving with AI are using it as a production tool for the parts of their practice that don't require their specific presence — background elements, colour iterations, quick mocks, grant applications, social content. The parts that require their specific presence, they protect.
Copyright: Genuinely complicated. AI-generated images have limited copyright protection in most jurisdictions — often none. If you're creating work for commercial use that involves AI, understand what you own. Adobe Firefly's training on licensed content is the most legally clear option for commercial work right now.
👇 Your exercise
Pick a project you actually have — a gig poster, an album cover concept, a product image, a grant application, social media content for your business. Use one AI creative tool to produce a genuine first draft.
Suggested path:
- If you need an image → start with Canva AI (zero friction) or Adobe Firefly (commercial-safe)
- If you need music → go to suno.ai, write a style prompt, generate 3 variations
- If you need writing → paste your grant brief or project description into Claude and ask for a first draft
- If you want to animate something → upload an image to pika.art and see what happens
The goal is one thing you can actually use or build on — not perfection, not an experiment for its own sake.
Branch complete 🎨
AI creative tools are production accelerants, not replacements. The ones who get the most out of them are the ones who already know what they're making — and use AI to make more of it.