StoryMaker

What it does

StoryMaker AI transforms spoken story ideas into complete animated short films. Users speak (or type) a story concept, and AI orchestrates a full cinematic pipeline: screenplay writing, storyboard illustration, video generation, narration, and multi-language translation all in one seamless flow.

How I used MeDo and the best part it generated

I used MeDo's Deep Build mode with 9 plugins orchestrated together in a single coherent pipeline - the deepest multi-plugin integration in this hackathon:

Plugin Pipeline (in execution order):

1. @Speech-to-Text - Captures voice via MediaRecorder API, sends audio blob to Whisper for transcription
2. @Large language Model (Gemini 2.5 Flash) - Expands the transcript into a structured 4-scene screenplay with visual descriptions, streamed via SSE
3. @Image Generation - Creates cinematic storyboard frames for each scene using the visual descriptions (async task with polling)
4. @Video Generation Omni (Kling) - Transforms each storyboard image into a 5-second cinematic video clip with camera movement
5. @Text-to-Speech - Generates professional voiceover narration for each scene
6. @Google Text translation - Offers the complete story in 5 languages (English, Spanish, French, Chinese, Hindi)
7. @Web Search + @The News - Powers a "Trending Inspiration" sidebar with real-time story prompts from current events
8. @Data Analysis - Provides "Story Intelligence" panel with word count, emotional arc, and pacing analysis

The most impressive feature MeDo helped create:

The streaming screenplay generation with real-time JSON parsing. MeDo generated a complete SSE (Server-Sent Events) streaming pipeline where the LLM response streams in real-time, the frontend parses JSON chunks progressively, and the UI updates live as the screenplay is being written. This creates a magical "watching the AI think" experience.

How I structured conversations with MeDo:

• Turn 1: Full architecture prompt with all 9 skills tagged, describing the complete app structure
• Turn 2: Fix prompt targeting the mic integration (replaced Web Speech API with MediaRecorder + Whisper pipeline)
• Turn 3: UX polish - added text fallback, recording timer, and error handling

Technical depth:

• Supabase Edge Functions as the backend layer, each wrapping a MeDo skill API
• SSE streaming for real-time LLM responses
• Async task polling for image and video generation (7-second intervals, 10-minute timeout)
• Proper error handling with quota detection (429), billing errors (402), and graceful fallbacks
• React Context for state management across the 5-page flow

What problem it solves

Content creators, educators, and marketers spend hours producing video content. StoryForge AI reduces "idea to finished video" from days to minutes. Speak your concept, get a professional animated short film with narration in any language.

Built With

• image-generation
• medo
• speech-to-text