Mateo and his friends sitting on the grass at sunset, a still from Mateo Vuelve

Animation studio. Spanish and English.

Short films that teach.

Animated films for children and the people who watch with them. One story, one lesson, made with care by people and our own production system.

Partner with us See the slate

First series: Mateo, a boy who plays soccer
Format: Short films, one lesson each
Languages: Spanish and English, voiced and subtitled
Status: In production

Why we make these films

Mamá comforts Mateo on the couch, a still from Mateo Vuelve

A good short film can stay with a child for years. We make ours calm enough to watch together and honest enough to talk about afterwards.

Education

Every film is built around one lesson sentence, written before the first scene. Anything that does not serve it is cut.

Sportsmanship

Sport is where children first meet winning, losing, teammates and rules. Our stories care more about how players treat each other than the score.

Curiosity

Patience and understanding other people. A child should finish a film knowing a little more about the world and themselves.

Feel-good

Warm, gentle and real. A character who loses is allowed to feel it, and every film ends with something worth smiling about.

The slate

Mateo. Season one.

Mateo learns about soccer, and about people, one short film at a time. It is our first series. The studio is built for many series and many sports.

FILM 05

Mateo Vuelve

Mateo Comes Back

Lesson: Getting back up
Spanish: Cuando te caes, te levantas.
English: When you fall, you get back up.

FILM 06

Ganar Bien

Winning Well

Lesson: Winning well
Spanish: Cuando ganas, ayudas al otro a levantarse.
English: When you win, you help the other team up.

Mateo about to kick the ball on the pitch

FILM 07

El Gol de Mateo

Mateo's Goal

Lesson: Everyone's goal
Spanish: Un gol es de todo el equipo.
English: A goal belongs to the whole team.

Mateo smiling beside the ball by a stone wall

FILM 08

Mateo practica

Mateo Practises

Lesson: Practise every day
Spanish: Si practicas todos los días, mejoras.
English: If you practise every day, you get better.

One lesson per film

Each film carries a single lesson sentence that the story earns.

Spanish and English

Titles, voices and subtitles in both, so families watch in the language they speak at home.

Made for children

Calm pacing, honest feelings and no tricks.

Many series, many sports

Characters, rules and sports live in data. A new series starts from its own series bible.

How we work

From one sentence to a finished film.

Ten steps, in order. Software does the repetitive work at each one. A person reviews the result and decides when the film moves on.

01
Idea
A short prompt and a one-sentence story. A person chooses which idea becomes a film.
02
Script
The screenplay, titles in both languages, and the lesson sentence.
03
Storyboard
Hand-drawn black-and-white panels, each with notes and its own approval.
04
Keyframes
Approved scenes become still images. Automatic checks flag problems; a person reviews every still.
05
Animate
Approved stills become short clips, following each scene's motion direction.
06
Voice
Every line is reviewed, then voiced in Spanish or English.
07
Music
The score is described in words, candidates are compared, one is chosen.
08
Edit
Scenes, storyboard, review notes and saved versions in one place.
09
Finalize
Rough cut, then the finished film with intro card, end card and subtitles.
10
Deliver
A private upload, or the film and subtitle files exported for a partner.

The coach kneels beside Mateo on the pitch

PANEL 04. APPROVED13 OF 13

The storyboard gate

Nothing is animated until a person approves the storyboard.

Every drawn panel is approved against the exact scene text and drawing. If the camera, action, cast, dialogue or length changes, that panel goes back to pending, and no final image work starts until every panel is approved again.

It is slower than skipping the step. It is also why each film says what we meant it to say.

Technology

Under the hood. Prompts, models, GPUs.

Each scene is a structured record. Software turns it into one text prompt, leases a GPU, and runs a model on our own hardware. Every result is checked by code and reviewed by a person before the next model touches it.

From prompt to picture

The still

01Inputs
- Scene record
  still camera faces head_count cast motion
- Series style
- Series rules
- Character descriptions
- A small, fixed number of reference images
02Prompt assemblyOne text prompt per scene, saved as a plain file.
03GPU leaseThe job asks a lease broker for a GPU under its own identity and priority.
04Image modelRenders the still on our own GPU. Drafts first, finals after approval.
05Automatic checksFaces, blank backgrounds, colours, small details such as glasses.
06Human reviewA person accepts the still or sends it back.

The shot

07Video modelAnimates the approved still, following the motion field.
08Speech modelVoices each reviewed line in Spanish or English.
09Music modelScores from a written description. A person picks one.
10AssemblyClips, voices, score, cards and subtitles in both languages.

The motion field is not part of the image prompt. It is read only when an approved still is animated.

Inference at work

Every image, clip, voice and score here is inference: finished models running on our own GPUs. The craft is the pipeline around them, so the same workflow runs on a small GPU or a large one, with whichever models are current.

A production run, as the scheduler sees itIllustrative snapshot. Machines are described by role only.

Machine	What it is doing	How it is scheduled
Machine A, image inferenceWorking	Draft stills: 10 of 33 done, now on scene 4. No errors in the log.	Holds a lease under its own job identity.
Machine B, audio inferenceIdle	Music and voices are finished. Nothing is running.	Both leases released.

Inference, not guesswork

Models run as finished tools. What we build is the pipeline around them: prompts assembled from records, leases, checks and human approvals.

Any GPU, any model

The workflow does not depend on one card or one model. A larger GPU changes how long a film takes, not how it is made.

Busy when there is work

Jobs take a lease, run, and release it when they finish, so a machine is working when there is work and idle when there is not.

What runs where

01Writing model

Offers story ideas from a short prompt. A person chooses one, or none.

02Image model

One assembled prompt plus reference images in, one still out. Finals reuse the chosen draft's prompt and seed.

03Video model

Turns an approved still into a short clip, directed by the scene's motion field.

04Speech model

Voices dialogue line by line, in Spanish or English, after each line is reviewed.

05Music model

Makes candidate scores from a description in words. Candidates are saved and compared.

06Where they run

On our own machines and our own GPUs. There is no cloud service in the production path.

07How they are scheduled

Every GPU job takes a lease with its own identity and priority, runs, and releases it. Our rule: nothing runs anonymously.

08What stays human

The lesson, storyboard approval, still review, the final cut and every upload.

A scene, as the software sees it

Illustrative. Field names are real, values are an example.

Scene record

still: A small grass pitch after rain. The goalkeeper dives low to her left as the ball rolls just past her gloves.
camera: low angle, from behind the goal, static
faces: goalkeeper focused, striker surprised and happy
head_count: 2
cast: ["goalkeeper", "striker", "ball"]
motion: Static camera. The keeper stretches, the ball rolls slowly into the net.

Assembled prompt

Warm stylised animated family-film look, soft morning light. A small grass pitch after rain. The goalkeeper dives low to her left as the ball rolls just past her gloves. Exactly 2 people in the frame. Camera: low angle, from behind the goal, static. Faces: goalkeeper focused, striker surprised and happy. Real green field, trees behind, no blank studio background. Goalkeeper: a girl of about eight in a green kit and orange gloves. Striker: a boy of about eight in a blue kit. Characters crisp, background slightly soft but readable.

Reference images attached: goalkeeper, striker.

Why this matters to a partner

Repeatable

The same scene record, style and rules produce the same prompt every time. A film can be rebuilt from its folder.

Auditable

Every storyboard approval is tied to the exact scene text and drawing. Films are saved as numbered versions with file checksums, never overwritten.

Replaceable parts

Models are roles, not commitments. We can change a model or add GPUs without changing how a film is made.

The studio

Our own software. Our own machines.

Estudio is a native production application and engine we built ourselves. Models do the heavy drawing, animation, voice and music work on our own GPUs. People write the stories, approve every step that matters, and own every file.

01ScriptWritten into the film file as structured scenes.
02StoryboardPasses only when every current panel is approved.
03RefsCharacter references keep faces consistent shot to shot.
04KeyframesDraft stills, then finals, each checked and reviewed.
05ClipsApproved stills animated into short clips.
06AudioVoices, the chosen score, subtitles in both languages.
07FinalAssembled with intro and end cards.
08UploadStarted only by a person.

People decide

A person writes the lesson, approves the storyboard, reviews every still and decides when a film is finished.

Local first

Stories, drawings and films are made and kept on our own machines and GPUs.

Honest storytelling

Real feelings, fair play, no tricks. We never make a child feel bad to make a point.

Portable by design

Everything a film is lives in plain files in one folder. Easy to back up, move and deliver.

Bilingual from the start

Spanish and English are part of every film, not an afterthought.

Open to change

Models are roles, not commitments. Better tools slot in without changing how we work.

Partner with us

Let's make the next one together.

Distributors, broadcasters, schools, sports organisations and investors: tell us who you are and what you have in mind. A person reads every message and answers it.

hola@estudio.icu