Story Creatures: Live

My daughter and I have a bedtime routine that's been going for years: she picks a story, I read it, and we talk about what happened for a few minutes before the light goes out. Some nights the conversation's short. Some nights we end up untangling something — why the fox gave up, why the mouse wasn't scared, why the tortoise kept going when everyone was laughing.

A few years in, I started to notice two things. First, the classic fables still land. A story Aesop wrote 2,500 years ago still helps a 4-year-old figure out her Tuesday. Second, the stuff available for kids online is either terrible (glittery, loud, designed to hijack attention) or expensive (the gentle stuff mostly costs money, and the best of it lives behind subscriptions).

So I started making them. One story at a time. Watercolored. Narrated in my own voice. Told like I'd tell it at bedtime — not the way an algorithm wants it told.

What Story Creatures is

Every story here is a classic public-domain fable, gently retold for modern kids ages 3–8. The art is generated with AI tools and then hand-edited for consistency — watercolor, warm light, no jump scares, nothing that looks like a video game. The narration is my voice, cloned so I can narrate 25 scenes in an evening instead of twenty-five. Each story comes with a free activity book — coloring, word searches, mazes, tracing — so the same story can live in your kitchen the next morning with a box of crayons.

What Story Creatures isn't

It isn't loud. It isn't flashy. There are no ads on this site. There are no ads in our videos (and there won't be — that's a line I'm not willing to cross for a preschool audience). We don't collect data on kids. We don't have an app that begs to be reopened. We don't have a character whose personality is "buy more of me."

We have stories. We have printables. We have a weekly email with one new story and one new activity book. That's the whole business.

People sometimes ask how one person makes a story a week. The short answer: I don't, exactly. I make the decisions, and a toolchain of small custom pipelines does the parts that don't require judgment.

A typical story takes:

• Script (human): I pick the source text, decide the retelling, break it into 20–30 scenes, and write the narration. This is the only part that matters; everything else is production.
• Audio (my cloned voice): ElevenLabs generates the narration in my voice, which I recorded once. It sounds like me because it is me.
• Illustration (AI, watercolor): A custom prompt per scene pushes a local image model (ComfyUI + DreamShaper XL) to produce soft watercolor art in a locked style. I review every scene and regenerate the ones that don't feel right.
• Activity book (code): A Python script reads the story's vocabulary and produces coloring pages, word searches, mazes, and tracing sheets tuned to that story. I review every book before it's offered.
• Assembly (code): FFmpeg stitches the video. ReportLab assembles the PDF. The site rebuilds itself when a new story lands.

It means the hard parts — the story, the voice, the review — stay human. The mechanical parts get faster. Which is how I can promise one new story every Tuesday without missing dinner.