AI Story-to-Storyboard Generator

Turn a full story into a scene-by-scene AI storyboard

The AI Story-to-Storyboard Generator takes a complete story or script and turns it into an organised, shot-by-shot production plan: character reference prompts, per-scene image prompts, minimalist image-to-video prompts with expected clip timing, and clean voiceover dialogue. Instead of writing every prompt by hand, you paste the story once and build the whole video plan in order. It works for any narrative video — short films, explainers, animation tests, and it ships with a ready-made preset for 1970s-style Sri Lankan folk-horror storytelling.

How it works (your key, your browser)

This tool has no server of its own. You connect your own AI provider — Google Gemini, OpenAI, or Anthropic — using an API key that is saved only in your browser (localStorage) and sent straight from your browser to the provider you choose. Nothing you paste is stored on this website. Because you bring your own key, you stay on your provider’s own free or paid tier and keep full control of your usage.

If you do not have a key yet, you can create a free Google Gemini key in Google AI Studio in about a minute. Paste it into the API key box, keep the provider on Google Gemini, and you are ready to generate.

Step-by-step: from story to storyboard

1. Add your API key. Open the settings panel, pick your provider, and paste your key. If a model error appears, set the model box to a current model (for Gemini, the free tier currently supports models such as gemini-3.5-flash).
2. Paste your full story or script into the main box. Separating scenes or beats with a blank line helps the tool split cleanly.
3. Run Part 0 — Setup. This generates every main character’s reference image prompt, a Key Objects & Props bible (recurring items like a lamp, a cloth, or ritual props), and a Location bible. This shared “bible” is what keeps characters and settings looking identical across every later shot.
4. Split the story into parts. Choose a part size in words — smaller parts give finer control, larger parts mean fewer steps — and the tool divides the story at sentence boundaries so nothing is cut mid-thought.
5. Generate each part. Generate parts one at a time, or use “Generate all remaining” to run them in sequence. Every part is shown next to its original story text so you always know which prompts belong to which passage.
6. Refine and export. Regenerate or clear any single part, pull all the dialogue into one voiceover block, build a master shot list, or download the whole storyboard as a file.

What each feature is for

Part 0, character and scene generators

Part 0 builds the whole world at once. If you only need one thing, the Character generator returns character reference prompts on their own, and the Custom scene box generates prompts for a single named moment (for example, “the old well at night”) without repeating the setup. Keeping these separate means each output stays focused and easy to copy.

Part-by-part generation

Long stories can overwhelm a single AI request, so the tool works in chapters. It carries the character and object bible forward from part to part, so a character introduced early keeps the same face, clothes, and props right to the end — consistency that is very hard to get when prompting scene by scene by hand.

Stable, editor-friendly prompts

The built-in instruction favours stable, repeatable shots: single clear subjects, one moving element per clip, close-ups over crowded wide shots, and implied action rather than complex motion. These choices dramatically reduce the warping and morphing that AI video tools produce on busy scenes, and they keep the footage clean and advertiser-friendly.

Voiceover, shot list, thumbnail and SEO tools

The Story tools panel adds finishing touches: extract every spoken line into one block ready for a text-to-speech voice, download the full storyboard as a Markdown file, build a numbered master shot list for your editor, generate a clean thumbnail image prompt, and produce platform-aware SEO — a title, description, and tags for YouTube, Facebook, TikTok, or Instagram, with the option to set your own focus keyword or let the tool propose one.

Your own instruction

The tool uses a built-in visual-director instruction by default, but you can upload your own .md or .txt instruction to change the house style, language, or output format. Your custom instruction is saved for next time, and there is a download link if you want the example as a starting point.

Tips for the best results

• Always run Part 0 first. It seeds the shared bible, so later parts stay consistent instead of re-inventing your characters.
• Pick a sensible part size. Around 400–800 words per part is a good balance of control and speed for most stories.
• Edit the running context. The context box is editable — tighten a character description there and every later part will follow it.
• Keep it advertiser-friendly. The default style implies dread through shadows and reactions rather than showing graphic content, which is safer for monetisation and gives cleaner AI results.
• Your work is saved. Everything stays in your browser until you press Clear, so you can build a long project across several days.

Troubleshooting

“This model is no longer available”

AI providers retire model names over time. If you see this, open the “See current models” link, copy a current model name into the Model box, and generate again. For a free Gemini key, try gemini-3.5-flash, gemini-3.6-flash, or gemini-3.5-flash-lite.

Nothing happens when I click Generate

You need an API key first. The tool will show a short set of steps and a link to create a free key. Add the key, then generate.

Frequently asked questions