Hey Product Hunt! 👋 VideoPrompt started with a simple frustration: every AI video platform wants a different prompt style, and "prompt generators" on the web all send your ideas and videos to their servers. So I built the opposite: • 100% on-device — real Qwen2.5 & Florence-2 models run in your browser via WebGPU. Your videos and images never leave your device. • 21 platforms, 100+ hand-written templates — Sora, Kling, Runway, Veo, Pika, Seedance and more, each with camera/lighting/style controls. • 5 local engines — text→prompt, video→prompt, image→prompt, prompt improver, and a character-lock card generator for AI short dramas (same face, same outfit in every shot). • 14 languages, free, no sign-up. And the part I'm proudest of: the entire template library is open source under CC-BY-4.0, with a pip CLI (vp-templates) and a public repo — because prompts shouldn't be locked in silos. Open DevTools: no keys, no analytics on the AI pages, no uploads. Built with WebGPU, WebLLM and Cloudflare Workers. Feedback welcome 👇

VideoPrompt

Free on-device AI video prompt generator — private

Free on-device AI video prompt generator — private