Sentimagem is KURACONV's three-pillar method for making cinematic AI films: Presença, Engenharia, Narrativa. It is the framework that lets the São Paulo studio direct AI instead of prompting it, shipping branded films of 15 to 89 seconds in two to three weeks that traditional production takes three to six months to make.

Most AI content is generated. A prompt goes in, a result comes out, and the maker accepts whatever the machine returns. Sentimagem rejects that posture. The method exists so that every decision is authored, not sampled, filtered through a council of 22 minds (photographers, directors, philosophers, strategists) before it reaches a frame. The principle is one line: we direct AI, we do not prompt it.

Pillar 1, Presença: the mood is the brand

Presença is coherence of mood across every asset. Not consistency of logo or font, coherence of feeling. The film, the editorial stills, the carousel and the original score must hold the same emotional temperature, so a viewer who sees only one of them still recognizes the whole.

In practice the mood is treated as the brand's primary asset. Every KURACONV project ships with an original composition, never stock, because a sound borrowed from a library breaks presença the instant it plays. São Paulo Blues, the studio's 2026 film, began as an original soundtrack: the mood arrived before the image.

Pillar 2, Engenharia: AI is the new lens, direction is the eye

Engenharia is camera direction in motion. AI is treated as a lens, a piece of optical equipment, and direction is the eye behind it. A lens does not decide the shot; the director does. This is where craft becomes engineering, controlled through hard production tooling rather than wishful prompting:

  • Shotlists that pre-commit every angle and movement before a single frame is generated
  • Identity locks that hold a face and form consistent across an entire film
  • Continuity filters that catch drift between frames before it compounds
  • Color grading and sound design applied as deliberate authorship, never presets

The toolchain, GPT-Image-2, Seedance, Higgsfield, Kling, Veo, is the equipment. None of it directs. That is the point: the model renders; the studio decides.

Pillar 3, Narrativa: story engine first, image last

Narrativa fixes the order of operations. Story comes first. Motion follows the story. The image emerges last, as the consequence of a narrative decision, never as the starting point. This inversion is what separates a film from a clip.

The studio's selected work runs this way. Dubai Falcon is a cinematic short built around a desert and a falcon. Everest is anchored by a single line, 'Who told you it was easy.' São Paulo Blues (2026) started as soundtrack and story before a frame was rendered. In each, the picture is the last thing decided, not the first.

Why the method exists

We direct AI. We do not prompt it.

Sentimagem is the operational form of 'direction over generation.' It is also a filter for what KURACONV refuses: no performance ads, no stock photography, no projects without a brief, no generic AI content. The studio takes a focused slate of projects, a deliberate constraint that protects all three pillars at once.

Founded in 2026 by Maria Rosa (creative director) and Diogo Felipe Silva (director), KURACONV STUDIOS is a borderless studio based in São Paulo and working globally. Its services run from editorial AI films and stills to carousels, brand identity, brand research and original music. Presença, Engenharia and Narrativa are how it keeps AI from collapsing into slop, by insisting that mood, direction and story are authored by people, and the machine merely renders the result.

Sentimagem holds that mood is the brand, AI is the lens and direction is the eye, and story comes first while the image emerges last.