Desktop video studio · every language

Your idea in.
A directed video out.

VoxForge researches, writes, finds the pictures, plans every beat, draws the maps and graphics, voices it and renders the final file. Not a template — a Director that decides every second, held to a quality gate before anything renders.

  • Ideas, scripts or recordings
  • Your own AI or ours
  • Renders on your machine
maps that move · rendered by the engine
300+coded graphics and devices
60transitions · 38 entrances
79font families, paired per video
RTL · CJK · Indicevery language, every visual

Three ways in. One Director.

Start from whatever you have. The same engine plans, checks and renders every kind of video.

💡

An idea

Give a topic. It researches with sources, writes a script with a hook that is scored and rewritten until it is strong, then finds real footage and pictures.

📝

A script

Your words stay yours. It plans every beat around them: maps, charts, cut-outs, moving words, camera and sound — each choice from a closed list, each placement by code.

🎥

A recording

It writes what to shoot and how to deliver each line, then locks your take to the plan: graphics beside you on the right word, never over your face.

Directed, not templated.

Every video gets its own look, its own type, its own telling — and the same check before it is allowed to render.

Maps that move

Armies advance and clash, heat flows from the sea, rain falls inside a country's real outline, cyclones turn — in your video's own colours.

Words on the beat

Every word lands on its spoken time, the key word hardest; split, gradient, outline and mirrored fills; 56 caption styles.

Plan check = quality gate

Before anything is built, the plan is measured with the same rules as the final check: reused clips, pace, pictures that really show the words.

Never the same twice

Fonts, colour harmony, 3D spaces and stagings are chosen per video. Same topic again? A different telling — unless you save a template.

Your own AI, per stage

Let our Director work, or hand research, planning and review to Claude Code, Codex, OpenCode or any MCP client — a different one per stage if you like.

Every language

Arabic, Urdu, Hebrew, Hindi, Chinese, Japanese, Thai and more — shaped text, bundled fonts, right-to-left layouts in every graphic.

Real footage, checked

Clips found on the web are judged against the exact words of their beat; cut-outs are made cleanly from found photos.

Private and resilient

Renders on your computer; a Wi-Fi drop pauses and resumes, a crash picks up from the last finished part. Your files never leave your machine.

Made by the engine

Thumbnails, chapters, titles and descriptions come with every render.

Thumbnail: One river, 240 million lives — the Indus drawn across Pakistan
The Indus: the river that made Pakistan14-minute documentary · web footage only
Thumbnail: Karachi is sinking
Why Karachi's sea is risingShort explainer · 3 numbers
Thumbnail: Split Pakistan — new provinces map
New provinces for Pakistan?Map-led explainer

How a video gets made

  1. ResearchSources, facts and the angle that has worked for others.
  2. Script & hookWritten for the first seconds; weak openings are rewritten, never kept silently.
  3. PicturesAsked for, generated or found — each checked against its words.
  4. The planEvery beat staged, dry-run against the gate, fixed until clean.
  5. RenderVoice, music, sound effects, a final check on the finished file.

Bring your own AI.

VoxForge speaks MCP. Connect Claude Code, Claude Desktop, Codex, OpenCode or any MCP client, choose which stage each one owns, and watch the work in the app. Or use any model by API key — Gemini, OpenRouter, local models.

Connect your AI →
// Settings → AI per stage
RESEARCH    → codex
DIRECT      → claude-code
PREVIEW_QA  → opencode
everything else → VoxForge

Simple pricing

Every plan has the same Director and every feature. Plans differ only in minutes of video.

Loading plans…

Compare plans

Questions

Do I need my own AI keys?

Not on the all-inclusive plan. On the other plans you bring a Gemini key (the free tier works) or any provider you like — or connect your own AI over MCP.

Does it render in the cloud?

It renders on your computer. A 4-core PC with 8 GB of memory works; 16 GB renders faster.

Will two videos look the same?

No. Looks, fonts, colours and stagings are chosen per video. If you want sameness — a channel identity — you save it as a template on purpose.

Which languages?

Every language the voice and model support, including right-to-left scripts and scripts without spaces, in captions, maps and every graphic.

Make the video you had in mind.

Download VoxForge