Pre-production
Before the shoot and before generation.
- Shot breakdown from the script
- 3D previs
- Camera blocking: position, movement, speed and lens
- Test shot of the hardest scene
AI film production · Feature films · Series
Professional AI film production for feature films and series
3D previs and camera blocking, generated shots, de-aging, voice conversion, visual dubbing and dialogue changes in post. We take on a single shot, a complete workstream or the entire film.
5,000+ videos delivered
In the business since 2017
4K master, EXR, DCP
The work we take on
We use generative tools for shots and tasks where a shoot or conventional VFX would be too expensive, too dangerous or no longer possible to arrange: destruction, crowds, backgrounds, a younger appearance, dialogue changed after the shoot, a foreign-language version.
The work fits into the production's existing post pipeline. We match generated shots to the filmed footage in colour, contrast and grain.
What we take on, by production stage
Visual effects
Shots for the finished film.
- Generated shots: destruction, crowds, battle scenes
- Backgrounds and set extensions
- Performance transfer from an actor to a CG character
Face and voice
Work on performances, with written consent.
- De-aging
- Facial modification
- Voice conversion: pronunciation correction, replacement of missing or unusable lines
Dialogue and localisation
After the shoot, without a reshoot.
- Dialogue changes in post, with lip sync
- Visual dubbing: lip movements matched to the foreign-language dialogue
End-to-end production
The entire film is made with generated imagery.
- From a human-written script and human performances
- With previs and camera blocking
- With post-production through to delivery
3D previs and camera blocking
Previs is a simple moving 3D sketch of a scene. It shows the layout of the space, where the characters stand, the camera move and the length of the scene. From it you can decide which shots to film and which to generate.
Camera blocking sets the camera position, movement, speed and lens for each shot. Generation starts from the approved previs and blocking. If you have a floor plan, set design or 3D model, we build the space from it.
How the work runs
Breakdown
We go through the script or the raw footage and define the tasks shot by shot.
Previs and camera blocking
Complex scenes get 3D previs and camera blocking.
Test shot
We start with a test of the hardest shot, and you can decide on the rest based on it.
Consent
We start work on a real person's face or voice only after obtaining their written consent.
Production
We make several versions of each shot and keep the one that matches the plan.
Integration and QC
We match the shot to the filmed footage in colour, contrast and grain, then check it frame by frame.
Approval and delivery
We deliver approved material as a 4K master, an EXR sequence or a DCP, to the production's specification.
Actors, rights and data handling
- Consent. A real person's face, likeness or voice goes into a film only with written consent.
- Actor models. A model built from an actor's face or voice is used only for that production.
- Model training. We do not use your script or footage to train models, and we upload them only to tools that do not learn from them either.
- Ownership and licences. Delivered and paid-for material belongs to the client (Terms). We hand over licence documentation for the models used when the project closes.
- Labelling. We label generated content as the law and the platforms require.
- Confidentiality. We sign an NDA on request. In that case the film does not appear publicly, by name or image, without your written permission.
We cannot name the titles of our feature film and series work. You can see more of our work in our showreel, and read about dubbing on our AI dubbing page.
Frequently asked questions
Frequently asked questions about AI film production
What is a quote based on?
The number and complexity of the shots, the material available and the deadline. Send us the script, a breakdown or the raw footage and we will send a tailored quote.
How long does it take?
That depends on the number of shots and the task. We set the schedule in the quote, based on your production's deadlines.
How do generated shots match the filmed footage?
We match them in colour, contrast and grain, then check them frame by frame. We start with a test of the hardest shot.
What formats do you deliver?
A 4K master, for example in ProRes, EXR sequences for the VFX team, or a DCP for theatrical screening. We agree the exact specification with the production.
Do you need the actor's consent?
Yes. We work on a real person's face or voice only with that person's written consent, and the model is used only for that production.
Will a model learn from our script?
No. We do not use your script or footage to train models, and we upload them only to tools that do not learn from them either.
Can you make an entire film with AI?
Yes. The script and the performances remain human work, and we generate the images from the approved previs and camera blocking.
Do you take on a single shot?
Yes. We can work on a single shot, one workstream or the entire film.
Which tools do you use?
We choose the tool for the job. For video generation we use Veo, Seedance, Kling or Runway, for example, sometimes several within one film.
Who owns the finished material?
Delivered and paid-for material belongs to the client, and we also hand over licence documentation for the models used. The copyright status of generated images can vary by country, so we agree this before production, based on your distribution plan.
Does generated imagery have to be labelled?
We label generated content as the law and the platforms require. What is needed depends on the content and the distribution, so we agree it before production.
Request a quote for your production
Tell us what film or series you are working on, which shots or tasks you need quoted and what the deadline is. We will get back to you within 24 hours.