Higgsfield Katana + Claude: 100% Code-Driven Product Launch Video Prompt

Product launch video production is among the most sought-after creative disciplines in modern motion design. Higgsfield AI (@higgsfield_ai) has published the full prompt and structural architecture for creating a high-end product launch film using Claude and Higgsfield Katana—relying on 0% video generation models and 100% code.

Image source: Higgsfield AI (@higgsfield_ai via X)

Instead of unpredictable video diffusion artifacts, this workflow relies on deterministic seeds, procedural graphic calculations, and separated photographic plates to build an editable, modular, code-driven motion composition.

Creator and Workflow Overview

• Original Creator: Higgsfield AI (@higgsfield_ai via X)
• Core Concept: 0% video generations, 100% code-driven launch film production
• Toolchain: Claude Motion + Higgsfield Katana
• Deliverable Specs: 30 seconds (24 fps, 720 frames total), 1080p main film and 2560×1536 showcase edition

Full Product Launch Video Prompt

Higgsfield AI structured the prompt into four distinct operational blocks (<inputs>, <direction>, <structure>, and <build>):

<inputs>
Use the attached PHYLLA video as the primary reference. Inspect its compositions, typography, scene changes, motion, and audio before building.
Ask me for any available original image assets, PHYLLA logo, brand fonts, Higgsfield/Katana branding, and soundtrack. Ask only for information that cannot be recovered from the reference.

If I skip anything, use these defaults:
Brand: PHYLLA.
Product descriptor: “NEW LANGUAGE MODEL”.
Ending tagline: “Generative intelligence”.
Duration: 30 seconds.
Frame rate: 24 fps.
Main film: 1920 × 1080.
Showcase version: 2560 × 1536.
Typography: a bold grotesk sans serif paired with a high-contrast italic serif.
Imagery: botanical macro photography, translucent biological forms, pastel sculptural objects, intimate human portraits, and procedural scientific graphics.
Use original assets where available. Otherwise, create suitable still-image replacements that match the reference’s composition, materials, and palette. Keep photographic assets separate from procedural graphics in the project.
</inputs>

<direction>
Create a complete launch film for PHYLLA using Higgsfield Katana and Claude Motion.
The story: a word behaves like a seed. A person supplies it, intelligence responds, and an entire world grows from that small beginning.
Starting palette: Near-black (#080A09), Charcoal (#141615), Ivory (#F4EBDD), Warm white (#EFEFEF), Pink (#E8A5C7), Peach (#F0BC9F), Mint (#BCDCCE), Muted green (#779B83), Higgsfield lime (#D1FF17).
Use bold sans serif for the main sentence and italic serif for the emotional or organic word.
</direction>

<structure>
Build seven chapters on one 720-frame timeline (30s at 24fps):
1. ONE WORD (0:00–0:04, frames 0–95)
2. SMALL BEGINNINGS (0:04–0:08, frames 96–191)
3. YOUR SEED (0:08–0:12, frames 192–287)
4. GROWTH (0:12–0:20, frames 288–479)
5. NEW WORLDS (0:20–0:24, frames 480–575)
6. PLANT A WORD (0:24–0:28, frames 576–671)
7. PHYLLA (0:28–0:30, frames 672–719)
</structure>

<build>
Build the film as an editable, code-driven composition using Katana and Claude Motion with deterministic seeds, procedural sphere arrays, and separated photographic plates. Export clean 1080p and 2560x1536 showcase.
</build>

Architectural Breakdown

1. Input Specifications and Default Parameters (<inputs>)

Directs the model to inspect compositions, typography, scene transitions, motion, and audio from reference media (the PHYLLA film). Explicit fallback defaults enforce strict brand naming, resolution variants (1080p / 2560×1536), and font pairing rules (bold grotesk sans serif paired with high-contrast italic serif).

2. Narrative Direction and Visual Palette (<direction>)

Defines the central metaphor: "a word behaves like a seed—a person supplies it, intelligence responds, and an entire world grows from that small beginning." The exact 9-color HEX palette (Near-black, Charcoal, Ivory, Warm white, Pink, Peach, Mint, Muted green, Higgsfield lime) guarantees strict visual continuity.

3. Seven-Chapter Timeline Structure (<structure>)

Splits the 720-frame (30s @ 24fps) timeline into seven chapters to eliminate timing drift:

1. ONE WORD (0:00–0:04, frames 0–95)
2. SMALL BEGINNINGS (0:04–0:08, frames 96–191)
3. YOUR SEED (0:08–0:12, frames 192–287)
4. GROWTH (0:12–0:20, frames 288–479)
5. NEW WORLDS (0:20–0:24, frames 480–575)
6. PLANT A WORD (0:24–0:28, frames 576–671)
7. PHYLLA (0:28–0:30, frames 672–719)

4. Build and Composition Rules (<build>)

Assembles the video as an editable code-driven project via Katana and Claude Motion using deterministic seeds, procedural sphere arrays, and cleanly separated photographic plates.

Original source

• Higgsfield AI Official X (@higgsfield_ai): Product launches is the one of the hottest skills on the market right now...