Styles: picking one, and blending three
The style field on the track card is labelled Clip style (1–3 presets, the first one is the base). That parenthesis is the whole lesson, but it deserves unpacking.
What a preset actually does
A preset is a written description of a look — optics, light, palette, texture, and a list of things the image must not contain. It is glued onto every frame prompt in the track. It does not describe *what* happens in the frame; that comes from the story, the comment and the scene text. A preset only decides how it looks.
Which is why the same storyboard in two different presets gives you two genuinely different clips, and why changing the preset halfway does not change frames that are already drawn.
The fifteen
Drawn and animated
- *Hayao Miyazaki (warm anime)* — hand-drawn, watercolour backgrounds, cosy light.
- *Cinematic anime (Shinkai)* — impossible skies, lens flares, emotional gradients.
- *3D cartoon (Pixar-style)* — glossy 3D, expressive faces, rich cinematic light.
- *Flat 2D animation* — flat vector, bold outlines, poster compositions.
Filmed
- *Realism (film)* — photoreal cinema, honest light, skin texture, a little grain.
- *Noir comic* — black and white with one accent colour, deep shadows, rain, neon.
- *DREAMCLAD (90s hood cinema)* — film grain, white tanks and bandanas, money,
doves and crosses, icon-like frontal framing.
- *FANUEL (cinematic surrealism, fire)* — a lone figure in a suit inside
impossible landscapes, fire everywhere.
Found footage and analogue surrealism
- *SPIKE (Russian cine-surrealism, cameos)* — panel blocks, old Ladas, smoke,
cartoon cameos played dead straight.
- *KATSUMI (found footage, surreal)* — rats, monks and aliens doing ordinary
human things, shot on a 90s camcorder with flash.
- *PUNKRF (found footage, Russian chaos)* — dashcams and VHS, red neon, absurdity
in the middle of traffic.
- *MUNIR (Gulf, flash, fisheye)* — street photography with flash and a fisheye:
rings pushed into the lens, a G63, Dobermans.
- *Long heads (90s analog surrealism)* — 90s film, surreal long-headed characters
living an ordinary street life.
Handmade
- *Claymation (plasticine)* — stop-motion, fingerprints in the clay, miniature sets.
- *Cardboard (thread embroidery)* — the whole frame stitched in thread on cream
felt and kraft board.
Why the analogue ones travel further
We pulled apart 640+ videos from five accounts that live on this exact aesthetic. The single most consistent trait across all of them: the generated image is disguised as an analogue one. Film grain, VHS, direct flash, fisheye. The brain reads the frame as a real photograph, and then the impossible thing inside it lands much harder — *wait, is that real?*
The other trait: one punch per frame, delivered deadpan. The frame reads in one second, without sound and without language, and nobody winks at the camera.
If you are making something to be posted rather than something to be admired, that is the direction: DREAMCLAD, KATSUMI, PUNKRF, MUNIR, Long heads. The polished 3D and the beautiful anime look better in isolation and travel worse.
Blending
Tick up to three. The first one you tick is the base — it carries the structure: optics, palette, how light behaves. The second one adds texture or subject matter. The third is seasoning, and often the reason a blend fails.
What works:
- A base plus a texture. *Realism (film)* + *Noir comic* — real cinema, pushed
into shadow and one accent colour. *DREAMCLAD* + *Noir comic* is the same move on a stronger base.
- Two neighbours from the same family. *KATSUMI* + *PUNKRF* are both found
footage; blending them widens the subject range without fighting over the look.
- A strong base plus a craft texture. *Claymation* + *Long heads* keeps the
analogue 90s framing and swaps the material.
What does not:
- Two complete worlds. *DREAMCLAD* + *MUNIR* both specify their own location,
palette, wardrobe and optics. They do not merge, they take turns — scene by scene, at random.
- Drawn plus photoreal. *Pixar-style* + *Realism (film)* averages into
plastic: neither honest skin nor honest animation.
- Three at once, most of the time. Three presets is roughly four hundred words
of style instruction competing for the same frame. Start with one, add a second only if you can say out loud what it is for.
One preset is a legitimate final answer. Most of the good clips we have seen use one or two.
Changing your mind
Switching the preset costs nothing. But if the scenes were already generated, the scene text was written under the old style — press Generate storyboard again before drawing frames, otherwise you get new-style pictures illustrating old-style directions.
Frames that are already drawn do not change on their own. Redraw them with ⟳ first, ⟳ last or Regenerate frames — free, once the scene has been paid for.
If you had written your own custom style text for the track, ticking a preset replaces it. The interface warns you: *this track has its own custom style — picking a preset will replace it.*