Module 4
Create with AI
The machines that make things — pictures, stories, characters. You'll learn how they actually dream up an image (real diffusion, step by step), how to steer them like a pro (prompts, seeds, guidance, control), and how to tell truth from fake and use it honestly. This is the real studio workflow: generate → control → verify. A 20-day journey ending in your own Honest AI Studio.
How to get the most out of this module
Generative AI isn't magic and it isn't a toy — it's the fastest-growing job skill in tech. This module teaches it the way a real studio would. Do these:
🎲 Part 1 · What Generative AI Really Does
It doesn't look up an answer — it rolls the dice and samples a possibility.
The Possibility Machine
Generative AI doesn't pick THE answer — it samples one of a million. Roll it and watch new stuff appear.
The Magic Seed
Same seed, same creature — every time. Meet the number that makes AI art repeatable.
The Weirdness Dial
Slide from safe-and-boring to wild-and-weird on a real language model running on your device.
Prompt = Recipe
A prompt isn't a spell, it's a spec: subject, style, mood, rules. Write one like a pro and watch it change the output.
Story Surgery
Make it shorter, funnier, calmer — with a real text-transforming model. Meet text-to-text generation.
🖼️ Part 2 · How Picture Machines Dream
The real mechanics of image generation — noise, denoising, latents, and control.
Pictures Into Words
A real model looks at your drawing and describes it. Vision + language, running on your device.
The Meaning Map
Words and pictures become dots on a map. Close dots = similar meaning. This is how prompts steer images.
Smash the Picture
Add noise, step by step, until a photo becomes pure static. This "forward" mess is step one of diffusion.
Un-Smash the Picture
Now run it backwards: remove a little noise each step and a picture appears from static. That's how AI paints.
The Guidance Dial
"More dragon, less banana." Turn up guidance to obey the prompt harder — and see where it breaks.
The Secret Small Canvas
AI dreams in a tiny compressed space, then blows it up. Morph one picture into another through it.
Erase & Redraw
Paint out part of a picture and regenerate just that patch. Meet inpainting and outpainting.
Boss the Picture Around
Give the generator an edge map or a pose to obey. This is how pros get exactly the image they want: ControlNet.
🎬 Part 3 · Real Studio Skills & Doing It Right
Consistency, tuning, checking your work, and creating honestly.
Same Hero, Every Panel
Build a 4-panel comic where the character stays the same. Consistency is engineered, not lucky.
Style Cards
Snap a small "style card" onto a frozen model to change its look — cheaply. Meet fine-tuning & LoRA.
Spot the Glitch
Six fingers, melted text, wrong count. Become an artifact detective — because generated media must be inspected.
The Judge's Table
Run a tournament: which output best follows the brief? Real models help score, but humans decide.
The Dataset Kitchen
What went INTO the model? Cook a dataset and watch your ingredient choices flip the copyright & consent risk.
Fakes, Watermarks & Passports
Spot a deepfake, stamp a content passport (C2PA), and learn why detectors are never 100% sure.
The Honest AI Studio
Make a real poster/comic with a prompt log, model card, evaluation, and an honest "made with AI" label.
Real generative tools & where to learn more
You know the ideas AND the pro words now. Here are the real, mostly-free tools creators use to generate, control, and verify AI media.
🎨 Make images & art
- Hugging Face Spaces — free in-browser image gen demos beginner
- ComfyUI — node-based pro image pipelines pro
- SD WebUI — run diffusion on your own GPU pro
🎛️ Control & fine-tune
- ControlNet — steer images with edges/pose pro
- 🧨 Diffusers — the real diffusion library pro
- transformers.js — models in the browser pro
🛂 Provenance & honesty
- Content Credentials (C2PA) — check who made it beginner
- C2PA standard — the provenance spec pro
- PAI Synthetic Media — responsible-use rules beginner
📚 Understand it deeper
- Diffusion Explainer — visual, interactive beginner
- Latent Diffusion paper — how SD works pro
- ControlNet paper — conditioning images pro