Petyo

Hey everyone, Arthur here, solo founder of Lugh Xtudio. We're launching today.

I built Xtudio because I wanted AI agents (Claude Code, Codex, Antigravity, or anything that speaks MCP) to edit on a real timeline, in my channel's style, and not just generate clips I can't touch afterwards. The launch video was edited entirely by agents inside the editor.

I'm curious how other creators see this. If you could hand off one part of your video production to an agent, which would it be: cutting silences, captions, B-roll, motion graphics, Shorts from long videos, or something else? And what's the part you'd never hand off?

Happy to answer anything about how it works under the hood.

I run SeedanceCheap. Studio 2 connects scene planning, video and image tools, Genjutsu transformations and multi-scene Productions, with one-time generation credit packs and no subscription.

Watch a complete 15-second Seedance 2.5 action take: an incoming punch, a trolley reversal and an intact-cake reveal.

This internal model test was requested at 720p and delivered at 720 1280 with generated audio. The complete 15.072-second original is shown without a recut or added upscaling. It is not a customer checkout or delivery test; some gestures and timing differ from the prompt.

I run SeedanceCheap. Studio 2 now connects scene planning, video and image tools, Genjutsu transformations and multi-scene Productions.

A three-scene structure you can try with any AI video tool:

1. Show an object that establishes a promise.

2. Introduce evidence that contradicts it.