# STUDIO CONNERS — VERTICAL VIDEO PRODUCTION STANDARD v0.2

**Updated:** September 17, 2026

## Purpose

Studio Conners turns selected Obsessions into outward-facing vertical video without creating a second content-production job.

The canonical pipeline is:

**Daily Obsession → ObsessOS Exhibit → Studio Conners Work Entry → Vertical Video Production Packet → production → distribution**

The video is a distribution artifact, not the canonical record. The durable record remains the ObsessOS Exhibit plus the Studio Conners Work Entry.

## Core production principle

**Standardize production, not imagination.**

Different Obsessions deserve different storytelling treatments, but every video should pass through the same editorial and production handoff.

Current video species:

1. **Cinematic Teaser / Artifact** — fictional worlds, striking objects, speculative concepts, atmosphere.
2. **Idea / Thought Experiment** — mechanisms, systems, experiments, intellectual possibilities.
3. **Build / Transformation** — products, workflows, websites, systems, before/after change.
4. **Studio Philosophy / Editorial Montage** — Studio Conners itself, its working philosophy, process, or operating system.

Real artifacts should remain real. Screenshots, interfaces, diagrams, objects, documents, and screen recordings should be preferred when they are stronger evidence than synthetic substitutes.

## Division of responsibility

### ChatGPT / editorial layer

ChatGPT determines:

- the strongest idea inside the Obsession
- whether the work deserves a vertical video
- video species
- hook
- story arc
- shot list
- narration versus on-screen text
- existing source assets
- assets that must be generated
- ending / payoff
- audio concept
- **audio-generation prompt**
- distribution packaging

The output is a standardized **Vertical Video Production Packet**.

### ComfyUI / generative production layer

ComfyUI may generate:

- still images
- image-to-video or reference-to-video motion
- visual connective material
- original instrumental music / audio beds
- sound design when an appropriate workflow is available

ComfyUI is a production engine, not the editorial brain.

### CapCut / deterministic assembly layer

CapCut or a similar editor handles:

- timeline assembly
- deterministic pans / pushes / crops
- typography
- transitions
- final audio placement and level
- fades
- final export

## Required Vertical Video Production Packet

Every Studio Conners vertical video packet should contain at least:

**Identity**
- Work Entry / Obsession
- video species
- target duration
- target aspect ratio and resolution

**Editorial**
- hook
- one-sentence story
- beat map
- narration, if any
- on-screen text
- ending / payoff

**Visual production**
- shot-by-shot timeline
- source assets
- generation prompts for visual assets
- motion instructions
- transition instructions
- end-card treatment

**Audio production — REQUIRED**
- audio role: music bed, sound design, ambience, narration support, or intentional silence
- **audio-generation prompt**
- target duration
- instrumental / vocals decision
- mood
- instrumentation / sonic palette
- tempo or energy level when useful
- structural arc across the video
- important sync points or beat changes tied to visual moments
- ending behavior: clean stop, decay, unresolved tail, sting, etc.
- negative constraints: what the track must avoid
- model / workflow used
- commercial-use / license note
- mix notes

**Distribution**
- primary cross-platform master
- platform-specific exceptions, if any
- channel copy / caption notes

## Canonical audio policy

The previous working rule was to export a silent video and add platform-native music separately. That created unnecessary platform dependence and inconsistent masters.

The new default is:

1. **Archive a silent master.**
2. **Generate an original audio track specifically for the video.**
3. **Create a finished Studio Conners audio master with that track baked in.**
4. **Use the same audio master across YouTube, X, Instagram, TikTok, the Studio Conners site, and other channels whenever platform rules allow.**
5. Platform-native music becomes an optional creative exception rather than the normal workflow.

This preserves one finished outward-facing artifact instead of rebuilding the soundtrack for every platform.

## Audio prompt requirement

Every video script / Production Packet must include an audio-generation prompt before production begins.

The prompt should describe sound directly rather than imitating a named artist or copyrighted song.

A strong prompt should specify:

- purpose of the track
- exact or approximate duration
- instrumental or vocal
- mood and emotional trajectory
- instruments / textures
- tempo or pulse
- density
- beginning, middle, and ending behavior
- moments that should align with important cuts or reveals
- things to avoid

### Audio prompt template

> Create an original **[duration]** instrumental audio bed for a vertical editorial video about **[subject]**. Begin **[opening character]**, develop toward **[middle movement]**, and reach **[payoff / reveal]** around **[timestamp]**. Use **[instrumentation / texture]** with **[tempo / energy]**. Keep the mix **[sparse / dense / restrained / cinematic / etc.]** so the visuals and any text remain dominant. End with **[ending behavior]**. Avoid **[vocals, genre cliches, specific sounds, excessive percussion, recognizable melodies, etc.]**. The result should feel original and should not imitate any named artist or existing song.

If the video intentionally needs silence, the packet should still include the audio field and explicitly state that **intentional silence is the audio decision**.

## Current audio-generation path

The current preferred experiment is **local audio generation inside ComfyUI** so the soundtrack can become another reproducible part of the Studio Conners production machine.

Current candidates include:

- **ACE-Step 1.5** for original text-to-music generation, especially instrumental beds.
- **Stable Audio Open** as an alternative for music, ambience, effects, and short audio generation.

Do not lock the system permanently to a model until the workflow has been tested on several Studio Conners videos.

For every model used commercially, preserve the model name, version, source, and applicable license terms with the production record.

## Audio generation workflow

Recommended default sequence:

**Production Packet → audio prompt → ComfyUI generation → select best take → trim / extend / regenerate as needed → CapCut mix → audio master**

Generate more than one candidate only when the first result is materially wrong. Avoid endless music iteration.

The audio should serve the story rather than become a separate creative project.

## Master files

Each completed video should ideally produce two masters:

### 1. Silent archival master

- 9:16
- 1080 × 1920 unless a later standard changes this
- 30 fps unless the production specifically requires another frame rate
- no licensed platform music
- no platform watermark

### 2. Studio Conners audio master

- same picture as silent master
- original, commercially cleared generated audio baked in
- audio mixed to support rather than overpower the visual story
- no platform watermark
- intended to be publishable across channels without modification

The **audio master is the default distribution asset**. The silent master remains the reusable source.

## Distribution rule

**Make once. Distribute many times.**

A video should not normally require a separate edit for every social platform.

Platform-native audio may still be used when:

- a specific trend or licensed track materially improves the piece
- a channel requires its own audio treatment
- the platform's commercial-use rules make the original master unsuitable

When this happens, the platform-specific version is an exception recorded in the Work Entry rather than the canonical master.

## Rights and originality guardrails

For generated audio used by Studio Conners:

- use a model / service whose terms permit the intended commercial use
- record the model and license in the production record
- do not prompt for imitation of a named living artist or a specific copyrighted recording
- do not use an existing song as a reference unless its rights clearly allow that use
- favor original instrumental beds and sound design
- keep the source prompt and final generated audio with the Work Entry or production archive

## First-video lesson that caused this revision

The first completed historical Studio Conners vertical video — **Bring ObsessOS Personal v0.1 to life** — was deliberately exported as a silent master. YouTube allowed music to be added after upload, while Instagram and TikTok pushed the workflow toward mobile editing and platform-specific music libraries, and X offered no comparable music picker.

That friction revealed a better system: Studio Conners should generate and own a commercially usable soundtrack as part of production, then distribute one finished audio master everywhere possible.

## Completion definition

A vertical-video production is complete when:

- the story is coherent at 9:16
- the visual master exists
- the required original audio prompt has been written
- the selected audio track has been generated and rights-checked
- the silent archival master exists
- the Studio Conners audio master exists
- the finished asset is attached to its Work Entry
- distribution can proceed without requiring the creative edit to be rebuilt per platform