Hi HN,
I recently pivoted a generic video-generation side-project into *DreamASMR*:
a web app that turns a text prompt into a fully-downloadable, 16:9 ASMR video with true binaural audio in ~30 seconds.
Why niche down to ASMR?
1. Slicing / pouring / tapping videos dominate “oddly satisfying” niches; audio quality is crucial yet underserved.
2. Veo3’s macro rendering + high FPS beats SD/Sora for tiny textures (glass, sand, honey).
3. Most existing tools lack a licensing path for commercial ASMR creators.
Key features
• *Prompt enhancer* – backend auto-adds acoustic & visual context so beginners get good results.
• *Binaural pipeline* – converts mono Kie.ai output to left-right stereo; best with headphones.
• *One-click download* – videos stored on Cloudflare R2; server proxy sidesteps CORS.
• *Credit billing* – only charge 1 credit on success, zero on failure.
Tech stack
• Next.js (App Router, mostly server components) + Tailwind + Shadcn
• Veo3 API via Kie.ai proxy
• PostgreSQL + drizzle-orm
• Cloudflare R2 for storage
• Next-Auth (Google / Email)
• Cloudflare Turnstile to stop bots
• 100 % TypeScript; minimal client-side JS
Demo
https://dreamasmr.com
Looking for feedback on
1. Generation quality – which objects/sounds still feel off?
2. Prompt UX – text box vs. template presets?
3. Pricing – would you pay for commercially licensed ASMR loops?
4. Perf / architecture – any bottlenecks in large file downloads or concurrency?
Roadmap
• Perfect-loop mode
• Pure audio ASMR generator
• WebM + alpha channel for compositing
• Public API / SDK
Thanks for reading & ship your thoughts!
I recently pivoted a generic video-generation side-project into *DreamASMR*: a web app that turns a text prompt into a fully-downloadable, 16:9 ASMR video with true binaural audio in ~30 seconds.
Why niche down to ASMR? 1. Slicing / pouring / tapping videos dominate “oddly satisfying” niches; audio quality is crucial yet underserved. 2. Veo3’s macro rendering + high FPS beats SD/Sora for tiny textures (glass, sand, honey). 3. Most existing tools lack a licensing path for commercial ASMR creators.
Key features • *Prompt enhancer* – backend auto-adds acoustic & visual context so beginners get good results. • *Binaural pipeline* – converts mono Kie.ai output to left-right stereo; best with headphones. • *One-click download* – videos stored on Cloudflare R2; server proxy sidesteps CORS. • *Credit billing* – only charge 1 credit on success, zero on failure.
Tech stack • Next.js (App Router, mostly server components) + Tailwind + Shadcn • Veo3 API via Kie.ai proxy • PostgreSQL + drizzle-orm • Cloudflare R2 for storage • Next-Auth (Google / Email) • Cloudflare Turnstile to stop bots • 100 % TypeScript; minimal client-side JS
Demo https://dreamasmr.com
Looking for feedback on 1. Generation quality – which objects/sounds still feel off? 2. Prompt UX – text box vs. template presets? 3. Pricing – would you pay for commercially licensed ASMR loops? 4. Perf / architecture – any bottlenecks in large file downloads or concurrency?
Roadmap • Perfect-loop mode • Pure audio ASMR generator • WebM + alpha channel for compositing • Public API / SDK
Thanks for reading & ship your thoughts!