Multimodal Studio

One connector. Image + video + consistent characters. Thirty-plus models, vendor-agnostic, brand-locked.

Purpose

Turn a creative brief into finished visual assets — images, videos, and reusable characters — without leaving the agent loop. Multimodal Studio is ACOS's unified generation layer: it routes each request to the right model, engineers the prompt, holds brand and character consistency across a whole asset set, and handles the async generation lifecycle end to end.

This is the capability that closes the gap with agent-first platforms like Google Antigravity — except ACOS stays model-agnostic (no lock-in to one vendor's image/video stack) and brand-aware (every asset inherits Frank DNA + project brand tokens).

When to Use This Skill

• Generating a hero image, thumbnail, OG card, or social visual for a piece of content
• Producing short-form video, b-roll, or a cinematic clip from a still or a prompt
• Creating a reusable character that stays consistent across many images and videos
• Building a coordinated asset set (article header + 3 social cards + a 5s teaser) that must look like one campaign
• Any time a workflow says ~~image generation or ~~video generation

The Connector (Higgsfield MCP — default)