A real-time image generator that actually streams pixels
Open-source frontend for Flux that shows images as they generate, not after.

What it does
BlinkShot is a Next.js web app that generates AI images through Together.ai’s API using the Flux Schnell model. The hook: it streams the generation process so you watch the image materialize rather than waiting for a finished file. It’s a reference implementation with analytics (Plausible) and observability (Helicone) wired in.
The interesting bit
The “real-time” claim is the product. Most image generators return a complete blob; this one appears to pipe progressive output to the browser. Whether that’s actual latent-diffusion streaming or clever polling of partial results is unclear from the README, but the effect is the same — users see something happening immediately.
Key highlights
- Streams image generation in-browser rather than blocking on full completion
- One-click deploy: standard Next.js app, single env var for Together AI key
- Flux Schnell via Together AI — fast, cheap inference, no local GPU needed
- Instrumented out of the box: Helicone for request tracing, Plausible for traffic
- Clean, minimal UI (Tailwind) with no auth barrier in the current build
Caveats
- Several obvious features are still TODO: no download button, no auth, no credit display, no gallery
- Rate limiting is currently IP-based, which is brittle for shared networks
- The “real-time” mechanism isn’t technically documented; could be streaming, could be rapid polling
Verdict
Good for developers who want a working Flux frontend to fork and extend, or who need to demo “live” AI generation to impatient users. Skip it if you need production auth, billing, or archival features today — the roadmap is still mostly unchecked.