Why I rewrote Leovidly in Rust

A while back I wrote that if I had to choose again, I'd still build Leovidly on Kotlin Multiplatform. I meant it at the time. I'm walking it back.

Here's what changed: I wanted to edit inside Leovidly.

The old Leovidly got you most of the way to a video — research, script, storyboard, assets, voice — and then handed you an export. To actually cut the thing, you left. You took your clips into a separate editor, learned its quirks, paid for it, and spent hours there. That last step always bothered me. The whole point of Leovidly is to keep the work in one place, and the slowest, most expensive part of making a video was happening somewhere else entirely.

So I set out to build the editor in. And that's where the old foundation ran out.

A real editor has to get close to the metal. Frame-accurate scrubbing, a timeline that doesn't stutter, a media pipeline you actually control, memory you can keep a lid on. The JVM and Compose Desktop are good at plenty, but they don't hand you the libraries or the low-level access to build that kind of editor well. I was already taping native pieces to the side — a Chromium browser through JCEF, FFmpeg for media, JavaFX for playback. Adding an editor on top meant more of the same, heavier, and fighting the runtime the whole way.

Rust gives me the access and the libraries the JVM couldn't. It gets me down to the level a performant editor needs, and the ecosystem around media, graphics and systems work is genuinely rich. Paired with a web front end for the interface, I get speed where it matters and room to move fast where I want it. The editor stops being a bolt-on and becomes part of the app.

That's Leovidly 2.0: rebuilt on Rust and Tauri.

Was rewriting a working app a big bet? Yes. I'm an Android engineer — Kotlin is home turf, Rust wasn't. Learning it while shipping was the cost. I think it buys something worth far more: one app where you make the video and cut it in the same place.