LOVE BEYOND PLANETS — A FIFTEEN-SECOND AI MICRO-CINEMA WORKFLOW

Updated: August 2, 2026

THE CONSTRAINT

Each Love Beyond Planets episode has fifteen seconds to establish three things: a world, a character action, and a change in feeling. The series avoids realistic synthetic humans on purpose. Six unfamiliar alien bodies make inconsistency less distracting and let gesture, scale, weather, and material become part of the emotional language.

The approved image remains clean: no burned-in captions, logos, or explanatory text. Accessibility and context belong in platform metadata or separate caption files, not inside the picture.

1. WRITE THE EMOTIONAL TURN BEFORE THE PROMPT

The unit of story is a three-beat change:

- Condition — what the character cannot do, reach, or understand.
- Recognition — what another being notices or changes.
- Release — the smallest visible proof that the relationship has moved.

For Episode 11, those beats were rejection, recognition, and release: Plu's signal is refused, Jupi calls Plu by name, and the signal finally leaves Pluto.

2. LOCK IDENTITY AND CONTINUITY IN STILL FRAMES

Each moving shot begins from a controlled 9:16 storyboard frame. The frame prompt describes recurring character traits that cannot change, the environment, prop geometry, emotional action, and prohibited artifacts.

For Plu, the lock included exactly five eyes, frost fur, three mismatched feet, the same signal box and antenna, and no added limbs, text, logo, or watermark. Identity is reviewed before motion generation because correcting a moving character after generation is less reliable than correcting the starting image.

3. GENERATE MOTION IN SMALL NARRATIVE UNITS

The recurring production pattern uses three approximately five-second image-to-video clips rather than asking one generation to invent the entire fifteen-second arc. Each clip receives one readable action and restrained camera direction. Cuts and editorial rhythm are decided by the human-led sequence.

Episode 11 used three Higgsfield Seedance 2.0 jobs, one per beat. The final edit trimmed each returned clip to five seconds and assembled them in narrative order. This is a documented episode-specific example, not a claim that every earlier episode used identical settings.

4. TREAT SOUND AS PART OF THE WORLD

The series does not use generic background music or human narration. Each planet is heard through weather, pressure, material, machinery, and short fictional alien vocalizations.

Episode 11 uses Pluto wind, brittle ice, a mechanical rejection clack, a low planetary resonance, and a clean signal-send tone. Small vocal sounds communicate attention and recognition without conventional dialogue.

5. EDIT, NORMALIZE, AND VERIFY LOCALLY

The release gate checks:

- exactly 15.000 seconds;
- full-frame 9:16 at 1080 × 1920 for the approved social master;
- H.264 video and AAC stereo audio;
- no replacement crop;
- no burned-in captions, background music, or human narration;
- character identity and story order at sampled frames;
- full decode without reported errors;
- a locked SHA-256 hash before upload.

The same approved file is uploaded once per platform. A public URL is recorded only after the native platform confirms publication; a planned time or attempted upload is not treated as a release.

6. SEPARATE CREATIVE LEARNING FROM DISTRIBUTION NOISE

The series measures each release at publication, +1 hour, +6 hours, +24 hours, and +48 hours. Platform metrics are interpreted by role: discovery on Instagram, view/like response on TikTok, Shorts-feed distribution on YouTube, qualified creative relationships on X, and attributed site behavior in GA4.

Re-uploading the same film, adding generic music, or rewriting every field during the observation window would make the creative lesson harder to interpret.

HUMAN-LED

- Series premise and character canon
- Episode writing and emotional structure
- Storyboard correction and shot selection
- Prompt constraints and continuity decisions
- Sequence, timing, sound direction, and final editorial judgment
- Release packaging, verification, and performance interpretation

AI-ASSISTED

- Generation of corrected storyboard imagery
- Image-to-video motion and native environmental audio
- Selected operational drafting and verification support

The result is not an autonomous film or an unedited model output. It is a small authored production system built around generative material.

DOCUMENTED EPISODE 11 CHAIN

- Corrected storyboard imagery: OpenAI built-in ImageGen; underlying model identifier was not exposed and is not inferred.
- Motion and native audio: Higgsfield Seedance 2.0, three five-second jobs.
- Local assembly and technical normalization: FFmpeg.
- Approved clean distribution master: episode-11-plu-your-signal-is-free-clean-1080p-v1.mp4.
- No third-party music, stock footage, celebrity likeness, trademarked character, or externally sourced dialogue was intentionally introduced.

This note describes the recorded production process. It does not make a legal warranty about platform terms, competition eligibility, ownership, or licensing. Tool attribution and applicable account terms must be rechecked for any festival submission, paid license, broadcast delivery, or partner warranty.

Project context: https://lovebeyondplanets.com/press
Press contact: beyondplanetlove@gmail.com