GlitchFlick

GlitchFlick turns any idea into a slick, creator-style video in minutes—just drop a prompt, pick a style, and get a script, memes, voiceover, and a ready-to-share video.

Comment

Inspiration

We love how short, witty, high-impact videos — like those from Fireship or Kurzgesagt — can teach or entertain in under 100 seconds. But making them is hard. Scriptwriting, voiceovers, visuals, editing — it’s a creative maze.

We asked: What if we could go from a single idea to a full video with memes, narration, and edits — using AI?

That’s how GlitchFlick was born.

What it does

GlitchFlick is an AI-powered web app that turns your video idea into a fully edited, creator-style video.

You just:

Enter a prompt

Pick your video style (e.g., Fireship, Humorous, Educational)

Get an auto-generated script

Curate meme-style images

Add an AI voiceover

Export a polished video in WebM or MP4

All in one flow — no editing skills required.

How we built it

Frontend: HTML + Tailwind CSS for modern, responsive UI

Script Generation: Google Gemini API (gemini-2.0-flash-lite) for fast, contextual scripts

Voiceover: Integrated ResponsiveVoice (fallback) and Google Cloud Text-to-Speech

Image Curation: Memes scraped via Imgflip API and Reddit’s meme stream

Video Export: In-browser canvas rendering and FFmpeg.js (planned) for compiling visuals + voiceover

UX: Step-based workflow with autosave, preview, image replacement, and voice tuning

Challenges we ran into

Gemini API’s limits on long outputs meant we had to tune prompts and parse scenes carefully

Voiceover syncing with visuals required experimenting with audio duration estimates

Browser-based video rendering (FFmpeg.js) is powerful but memory-intensive — we needed fallbacks

Managing different APIs (Gemini, voice, memes) and asynchronous tasks in a smooth UX flow

Accomplishments that we're proud of

Fully functional prompt-to-video pipeline

Script formatting with scene parsing + meme relevance

Polished interface that’s usable even on mobile

Dynamic voice preview, editing, and video-style flexibility

0-install, web-based — accessible to anyone with a browser

What we learned

How to prompt-engineer creative outputs using Gemini

Audio/visual syncing without a full video engine

Real-world API integration (TTS, memes, images) with failover handling

UX matters — guiding the user through creativity without overwhelming them

What's next for GlitchFlick

🎥 Backend rendering: Offload FFmpeg to server for faster, more reliable exports

🧠 Voice cloning: Let users generate content in their own voice

🌍 Community gallery: Share your GlitchFlicks with the world

🛒 Pro features: Templates, auto thumbnail generation, YouTube upload integration

🤝 Collab mode: Script + meme editing with friends in real-time

Share this project:

Updates