Open Studio

Title in · Script in · Finished video out

Video Production Engine.

Seven short-form formats, three ways to build them: licensed stock footage, AI generation, or both. ShoAI researches your script, plans the scenes, sources the footage, voices the narration, burns the subtitles and renders the master — automatically.

No editing timeline · No render farm · A title and a script is the whole brief

Seven bars, seven formats — tap one

1080pH.264 master + SRT subtitles
~3 minScript to finished master
3Aspect ratios: 9:16 · 16:9 · 1:1
100%Licensed stock footage
Three ways to build it

The bars are what.
This is how.

Any format can be assembled three ways. The pipeline picks the sensible default for every scene — and you can force a mode when you know what you want.

Mode · Stock

Licensed Stock

Real footage from licensed stock libraries, matched to your script scene by scene.

  • Every clip carries a confirmed licence — no scraping, ever
  • Orientation-matched to your aspect ratio
  • Real people, real places, zero uncanny valley
  • The fastest path to a finished master
SOURCE · PEXELS API SCENE COST · FREE
Mode · AI

AI Generated

Fully generated clips when the shot you need doesn't exist in any library.

  • Text-to-video per scene from the planned visual prompt
  • Impossible locations, abstract visuals, product mockups
  • Serverless GPU backends — nothing to install
  • Switch on with one environment flag
BACKENDS · LTX / REPLICATE / HF TOGGLE · FORCE_AI_VIDEO_GENERATION
Mode · Hybrid

Hybrid

Stock where it lands, generation where the library runs dry. The default.

  • Research flags the scenes stock can't cover
  • Generation fills only those gaps
  • One consistent grade across both
  • Best quality-per-minute of the three
SELECTION · PER SCENE FALLBACK CHAIN · AUTOMATIC

Whatever the mode, the rest of the pipeline is identical: AI voiceover, phrase-timed subtitles, background music bed, per-scene pacing locked to the narration, and an FFmpeg master with your title burned in.

The bin — formats it ships

Pick a format.
See what it makes.

Generate one like this Title + script in, master + subtitles out. Watch it move through the pipeline live.
How it runs

Script in.
Master out.

STAGE 01

Script

You give it a title and a narration script. Pick duration, aspect ratio and style. That's the whole brief.

→ 1 MINUTE OF TYPING
STAGE 02

Research

Keywords are extracted, facts are pulled from live web search, and licensed footage is matched to every scene.

→ ~30 SECONDS
STAGE 03

Assemble

Scenes are planned and paced to the voiceover, narration is synthesized, subtitles are timed phrase by phrase.

→ ~90 SECONDS
STAGE 04

Deliver

FFmpeg burns the title and subtitles and encodes the master. Watch it in the studio or download the MP4 + SRT.

→ SAME MINUTE
Honest comparison

Against editing
it yourself.

Where the pipeline wins, and where it doesn't. If your job is in the last row, open a timeline editor — that's the right tool.

 ShoAIManual editing
First cut~3 minutesAn afternoon, if it goes well
Footage sourcingLicensed, automatic, per sceneTab-hopping stock sites for hours
VoiceoverSynthesized with the masterMic, takes, retakes
SubtitlesPhrase-timed, burned inTyped by hand, drift by minute two
Ten variants of one ideaTen scripts, ten jobsTen timelines
Frame-exact art directionNot this tool's jobOpen the editor
Get started · Two minutes

Bring a script.
Leave with a video.

The studio is self-serve. Make an account, paste a script, and watch the job move through research, planning, assembly and render — live, stage by stage.

How it works

  1. Create an accountEmail and a password. A workspace is provisioned automatically.
  2. Start a new videoTitle, script, duration, aspect ratio, style. The form is the brief.
  3. Watch the pipelineEvery stage reports live: research, scene planning, assets, render.
  4. Review the masterPlay it in the studio with subtitles, straight from the job page.
  5. Download and shipMP4 master plus the SRT file, ready for any platform.

What a job gives you

DeliverableSpec
Video master1080p H.264
Vertical 1080×1920, landscape or square
VoiceoverSynthesized
SubtitlesBurned + SRT
Music bedMixed at −18 dB
Title cardBurned in
Fades in over the opening scene

Four rules for a script that lands

  • Put the hook in the first sentence — the scroll decides in two seconds.
  • One idea per sentence. The planner cuts scenes at your full stops.
  • Concrete nouns make better footage matches than abstractions.
  • 50 characters minimum, 2,000 maximum. Aim for the duration you picked.
Answers

The questions
you'd ask anyway.

What exactly does ShoAI do?

It turns a title and a narration script into a finished short-form video. The pipeline extracts keywords, researches facts from live web search, plans a scene-by-scene timeline, matches licensed stock footage to each scene, synthesizes the voiceover, times subtitles phrase by phrase, mixes a music bed and renders a 1080p H.264 master with your title burned in. You watch every stage happen live on the job page.

How long does one video take?

Around three minutes from pressing Generate to a playable master on a standard 15–30 second brief. Research is roughly half a minute, assembly and voiceover about ninety seconds, and the FFmpeg render lands in the same minute. Free-tier API rate limits can add retries; the pipeline waits them out automatically.

Where does the footage come from?

Licensed stock libraries, fetched through their official APIs — every clip arrives with a confirmed licence and is matched to your project's aspect ratio. The pipeline never scrapes the web for video. When a scene can't be covered by the library, AI generation can fill the gap through serverless backends.

Is the video fully AI-generated?

By default, no — the footage is real licensed stock, which is why the output doesn't have the uncanny "generated" look. The AI does the editorial work: research, scene planning, footage matching, voiceover, subtitle timing and assembly. If you want fully generated visuals, a single environment flag switches every scene to text-to-video generation.

What formats and ratios do I get?

1080p H.264 masters in 9:16 vertical (1080×1920) for Reels, Shorts and TikTok, 16:9 for YouTube, or 1:1 for feed — chosen per project. Every job also delivers the SRT subtitle file alongside the burned-in captions, and durations run 15, 30, 60 or 90 seconds.

Who owns the output?

You do. The stock footage is licensed for use through the provider's API terms, the voiceover and edit are generated on your account, and the master lands in your workspace's storage. Download it and publish it wherever you like.

What does it cost to run?

The platform is self-hosted — FastAPI, Celery, Postgres, Redis and Next.js in Docker. The only external costs are the API keys you plug in, and every provider in the default configuration has a workable free tier. A cost ledger tracks estimated spend per stage of every job so nothing surprises you.

How do I start?

Create an account, press New Video, and paste a title and a script. A workspace is provisioned automatically on sign-up, the form is the whole brief, and your first master is about three minutes away.

Roll
camera.

Paste the script. Watch the pipeline build your video.

Open the studio
00:00:00:00
00:02:30:00