From an idea to a finished film, in one place.

An AI film studio in the browser. From an idea or your own screenplay it writes the story, creates characters, settings and props, draws the storyboard, generates every shot with video and sound, narrates, scores and edits it, ready to export, publish on social networks or carry on in Unreal Engine and Unity. It is made by Utopia Dreams VFX, Raúl Bellmunt Tabero's studio in Terrassa (Barcelona, Spain), a member of the Google for Startups Cloud Program. It is in beta: you request access and the team approves it.

1) Write a pitch or import your screenplay. 2) The app writes the story and creates the cast, settings and props with images. 3) It draws the storyboard shot by shot. 4) It films each shot with video and sound, first as a cheap draft, then in final quality. 5) Add narration, dialogue with fixed or cloned voices, music and lower thirds. 6) Edit, give it a film finish (grain, vignette and colour, at no credit cost) and export as MP4, for social networks, as SCORM or to a game engine.

Video generators make loose clips and avatar platforms make a talking head. Utopia carries the whole film: script, a cast that keeps its face in every shot, storyboard, shots with camera and sound, dubbing into other languages with the same picture, editing, SCORM packages for training, previs in Unreal Engine and Unity and backgrounds for virtual production LED volumes.

It does a different part of the job. Google Flow and Runway generate and string together AI video shots; Synthesia and HeyGen make videos with an avatar talking to camera. Utopia uses video models such as Veo, the one behind Google Flow, and builds the whole film around them: script, a cast with the same face in every shot, storyboard, shots with sound, narration, music, editing and delivery as MP4, as SCORM or to Unreal Engine and Unity. If you only need a single clip or a presenter reading a text, those tools are more direct; if you need a piece with a story, several characters and an edit, Utopia saves you moving from one to another.

Any audiovisual piece with a story: ads and brand videos in several formats and languages; training videos with a presenter, subtitles and a SCORM package, also in batches with the Series add-on; short films and fiction; storyboards and previs to prepare a live-action shoot, which carries on in Unreal Engine or Unity; and backgrounds for LED virtual production stages. There is a Beginner mode for newcomers, and a Filmmaker mode and team spaces for production companies.

Each character's photo is the identity reference in every image and shot, and a fixed or cloned voice says their lines. On top of that, the coherence advisor checks each step against the continuity rules (photos, voices, clothes, faces in the first frame, camera, light), says how to fix each warning and fixes what it can in one click, before the clips are paid for. Its rule checks use no credits.

The interface is in English, Spanish and Catalan. Stories, voices and subtitles can be made in more than 30 languages, Catalan, Galician and Basque included, with Antton, a native Basque narrator, on the Pro and Estudio plans. A film can be dubbed into another language keeping the picture.

Yes, over MCP, the open protocol for AI tools. Claude.ai, Claude Desktop and ChatGPT connect by signing in to Utopia; Cursor, Claude Code, Gemini CLI, VS Code, Codex CLI and Devin Desktop use an access token from Settings → «AI assistants (MCP)». The assistant can read your stories, create new ones, write their characters, settings, props and shots, draw their pictures (it tells you the credits first), bring back an earlier version, export the movie, change the edit and send orders to Unreal Engine or Unity. It comes with the Estudio plan.

There is a free plan with 300 credits a month and three paid plans: Creador (€49 a month), Pro (€149 a month) and Estudio (€599 a month), which make about 2.2, 7 and 28.2 minutes of finished film a month, 17 % off when paid yearly. Each paid plan starts with a free trial of 18 days, once per person, which you can cancel in one click before the first charge. Usage credits that never expire can also be bought, and the models from other companies (not Google) come in an add-on for the paid plans, one payment of €29. Each AI call spends credits by model and quality, and drafts cost far less than the final take. Prices without VAT.

Yes. The Free plan gives credits every month without a card, enough to try a draft, and each paid plan starts with a free trial you can cancel before the first charge. During the beta, access is by request: ask for it on this page and an administrator approves it.

Each AI call (text, image, video, voice or music) spends credits by model and quality. You first film each shot as a cheap draft and only make the final take of the shots you approve. The plan cards show how many minutes of film and how many ads, shorts or training videos each plan makes.

Yes. The help assistant (Help → Ask for help) answers in the app at any time, and with a support package you open a request to the team from the same chat and follow it in «My requests». On top of any plan, Free included, you can add Basic (€19 a month, first answer within 2 working days, help with your account and invoices, video guides) or Premium (€129 a month, priority requests answered within 8 working hours, a named person on the team, a monthly written review of your stories and help with Unreal Engine, Unity and LED walls). Support is by request only, not by phone, and response times are targets, not guarantees.

Google's latest models on Vertex AI: Gemini for the script and the assistants, Nano Banana for the images of characters, settings and storyboard, Veo 3.1 for video with sound, Lyria for music and Gemini Text-to-Speech for voices. Stories in Basque can be narrated by Antton, HiTZ's open Basque voice, which runs on Utopia's own servers. With an add-on, the paid plans also get video and 3D models from other companies: MiniMax, Kling, Wan, Seedance, Hunyuan3D, Trellis and Tripo. The AI models page lists each one and what it is best at.

Yes. Beginner mode picks the most efficient models for you; Professional and Filmmaker let you choose each one, from Veo 3.1 Lite for drafts to Veo 3.1 for the best take, in 720p, 1080p or 4K, or, with the add-on, another company's video model, and the app tells you the credits before you spend them.

You do. What you upload or write is yours and, as far as the law allows, so is what the app generates for you, commercial use included. AI models may produce similar results for other people, and generated content may not be protected by copyright in every country.

No. Your content is only stored and processed to provide the app. Google's models run on Google Cloud's Vertex AI, whose terms don't let Google use it to train its models, and open models such as the Basque narrator run on our own servers. The other companies' video and 3D models and Sync Labs' lip sync run on fal.ai only when you use them, and never if you choose «Google models only» or «Data in the EU». The privacy policy lists every provider and what each one receives.

Yes, as article 50 of the EU AI Act requires: every movie and soundtrack you export says in its metadata (IPTC standard) that it was made with AI, the images, clips, music and voices from Google's models carry Google's invisible SynthID watermark, and shared films, YouTube descriptions and social posts say so too. You can also put the EU's «AI» label on the picture when you export.

If you publish for your work or business, you tell your audience it's AI. A film with realistic people, places or events (an advert, a training or informative video) needs the «AI» label where viewers see it: in the editor, Export → «AI» label on the picture → The whole film. In an obvious work of fiction, the start or a mention in the credits or the description is enough. Don't remove the marks from the files, and label AI-written text that informs the public on matters of public interest unless a person has reviewed it.

Yes. The Unreal plugin and the Unity package open the story with its shots, cameras, characters and sound, and changes come back to the app. It also makes backgrounds for virtual production LED volumes.

Yes. It makes seamless looping backgrounds for virtual production LED volumes: you first review a keyframe every 8 seconds, then Veo films each segment and your browser joins them into an MP4 that loops without a jump, as a 720p draft or a 4K final.

Yes. It has training templates (case study, good and bad practice, micro-lesson, presenter to camera), a lip-synced presenter, lower thirds, subtitles in several languages and SCORM 1.2 and 2004 export with a transcript, ready for any learning platform (LMS). A scene with long dialogue can be one continuous shot of up to 40 s, with every word said in the characters' voices. It also produces series of videos in batches from a spreadsheet.

Yes. A team shares its stories and a common wallet of credits that the owner manages. Invite people by email, bring a story you started alone into the team, and set up a Google Meet with everyone from the team window. Each paid plan says how many people its team takes.

Yes. With the Series add-on, shown next to the plans, you create your own series and invite your people: download the template from the series page and fill in one row per video, and the Utopia team prepares the production and makes it in batches with the same presenter, voice, opening and closing. Scripts can have scenes where characters talk to each other, and a shot lasts up to 16 s, cut only between sentences. In a review queue you approve each video before it is delivered as MP4, SCORM and subtitles. For a large series, write to hello@utopiadreams-vfx.com.