Features

One studio to finish a whole film.

Write the script, split it into shots, generate keyframes and video, arrange the timeline and export MP4 — all in Nomi. You bring the models and pay the provider’s price; each project is a folder on your own disk.

Agent

Say what you want, and it gets to work

Describe the film in plain words. The agent reads your canvas, splits the story into shots, generates keyframes and video and lays them on the timeline. Everything it makes lands on the canvas, where you can see it and change it.

Understands what is already on the canvas
Shots, prompts, images, video and timeline in one pass
Backed by directing and writing skills, so shots are planned

See the skills it uses →

You decide

You choose when it asks

Three levels: ask at every step, edit automatically, or fully automatic. On the default level it edits the script and timeline without interrupting you, but still asks before any paid generation. Full autonomy needs your explicit confirmation, and irreversible actions always ask first.

Default level asks before spending money
Full autonomy has to be switched on by hand
Irreversible actions such as deleting always ask

Storyboard

One row per shot, and you direct

Each shot is one row: prompt, first frame, model and duration at a glance. Lock a reference card for a character or location and every shot that uses it keeps the same face and the same place.

Edit prompts and switch models row by row
Reference cards keep characters and places consistent
Pick a model per shot

Character and scene recipes →

Generation canvas

Every take is kept, so you pick the best

Shots become nodes on the canvas. Switch between text to video, first frame, first and last frame, and multi-reference. Every generation stays as a version; if it is not right, make another and put the best one on the timeline.

Four modes: text, first frame, first and last frame, multi-reference
Compare versions of the same shot side by side
Apply effect prompts to a card with one click

Browse the prompts →

Every take is kept, so you pick the best

3D director’s stage

Your phone is the viewfinder

Pose the characters and place the camera in a 3D scene, then frame the shot with your phone. The frame goes straight back to the canvas and on into video.

Pose characters and place cameras
Scan a code and use your phone as the viewfinder
Use the captured frame as the first frame

Any model

Use any provider, switch per shot

Try the same shot on Seedance, Kling, Wan or Hailuo. Connect APIMart, Kie.ai, Volcano Engine Ark, ModelScope, a Jimeng membership, any OpenAI-compatible relay or a local ComfyUI; ask Claude Code or Codex to connect a new model for you.

Your own keys, at the provider’s price
The model picker shows where each model is available
New models can be connected the day they launch

Browse the models →

Timeline and export

Arrange it, add sound, export MP4

Put the chosen shots in order on the timeline, add captions and music, preview the whole film and export MP4.

Image and video tracks, music and captions
Preview while you edit
Export MP4 in one click

Read the getting-started guide →

What it can’t do yet

Precise lip sync is not supported yet.
The Mac app is not signed by Apple yet: open it with right-click → Open the first time, and updates need a fresh download.

Everything stays yours

No account, no telemetry. A project is a folder on your disk with the canvas, storyboard, timeline and every image and clip inside. All of the code is open source (AGPL-3.0).

Pro-grade AI video. Models at their real price.

macOS · Windows · AGPL-3.0 · No account

Download Nomi