🎬 Omni Bridge
Multimodal router · free-first · sidecar orchestration
Video Router

Generate through Omni without breaking your existing model flow.

Omni Bridge is the C2C sidecar orchestration lane for multimodal video work: prompt in, references in, route selection, conversational revision, asset packaging, and fallback handling.

Current structure supports text, image, audio, and video inputs; up to 5 reference photos; avatar-aware jobs; 10-second clip planning; and fallback routing across Gemini App, Flow, YouTube Shorts, and future API paths.

Routing defaults
  • Preferred lane: Gemini App
  • Fallback chain: Flow → YouTube Shorts → Vertex API
  • Mode: Free-first with provider-aware fallback
  • Output pack: Video, thumbnails, captions, prompt log, publish payload
Inputs
Prompt · images · video · audio · brand assets
Consistency
Up to 5 refs · identity notes · continuity rules
Edit memory
Structured conversational revision history
Workflow chain

Canonical local-service engine currently modeled as structured stages rather than direct provider lock-in.

Lead sourceraw leads
Qualificationfit + niche score
Prompt packprimary + alt prompts
Sample videodrafts + thumbs + captions
Outreach offermessage sequence + package
Scheduler payloadpublishing assets
Retainer templatemonthly plan + pricing
Prompt object

Use a structured job shape with intent, inputs, references, avatar, routing, constraints, and outputs.

Current guardrails

Treat short clips as chained scenes, preserve watermark awareness, and keep route selection separate from creative intent.

Next build

Wire this page to load config JSON dynamically and turn it into a real orchestration cockpit.

Visual proof

These mock sections turn Omni from a text-only concept into a believable routing product with visible workflow evidence.

Routed job mockup
Prompt
Create a cinematic 10-second promo for a local luxury barber brand using 3 reference photos, dramatic light, premium closeups, voice-led CTA.
Reference photos
Founder portrait Barbershop interior Product closeup Chair angle Street signage
Edit mode
Generate Avatar on 10s Native audio
Fallback chain
Gemini App → Flow → YouTube Shorts → Vertex API
Route summaryFree-first mode
Primary lane
Gemini App
Continuity rules
Keep face, outfit palette, shop interior angle, gold trimmer highlights.
Output packReady for publish
MP4 draft Captions Thumbnails Prompt log Publish payload
Reference consistency

Show how identity, setting, and product continuity can hold across generations and edits.

Identity hold
Face shape, beard line, and jacket silhouette preserved.
Location hold
Interior tone and mirror wall remain consistent.
Wardrobe hold
Color palette and styling stay locked between scenes.
Product hold
Hero trimmer shape and finish remain recognizable.
Look hold
Lighting language remains premium after revisions.
Generate → Edit → Extend

A simple storyboard rail makes Omni’s workflow legible at a glance.

Step 1
Generate
Start with the initial scene using prompt, reference photos, and routing preferences.
Step 2
Edit
Apply conversational revisions while preserving the subject, setting, and visual continuity.
Step 3
Extend
Continue the sequence into the next beat without rebuilding the whole creative direction.
Status

Omni Bridge is now a sidecar workflow lane, not a replacement for your existing system. The current Studio structure can keep its normal free flow while Omni evolves into a reusable orchestration engine on top of it.

Important: keep the live file and canonical source file synced after every patch.

\n \n

Omni Bridge · I/O 2026 Schema v2026.05.24

abel>Prompt abel>Reference Photos (up to 5)
abel>Edit Mode
abel>Duration (s, max 10)
abel>Avatar ID (optional)
abel>Audio Reference (optional)
ready · routes Gemini App → Flow → YT Shorts → Vertex API (when live)