odyssey-interactive-video
Plan and scope production work with Odyssey's real-time interactive video world models (odyssey.ml — Odyssey-1/Odyssey-2/Starchild-1/Agora-1). Use when a request involves generating video that a user steers live with keyboard, text, speech, or controller input (streamed frames that respond in the moment), or when someone must decide between Odyssey-style interactive generation and offline clip generators (Sora/Veo/Kling) or downloadable 3D / game engines. Covers what is verified versus announced, current access routes, latency and coherence limits, feasible-now versus speculative use cases,
Profile →npx skills add calesthio/generative-media-skills --skill odyssey-interactive-video --agent claude-code
Same command for any agent — swap --agent for codex, cursor, copilot.
Weekly change comes from our own snapshots, not the repository page — it measures attention, not adoption.
What it does
Plans and scopes production work for Odyssey's real-time interactive video world models, guiding when to use live-steered video, choosing between Odyssey-style interactive generation and offline generators or downloadable 3D/game engines, and outlining verification status, access routes, latency/coherence limits, pricing/access terms, and content/rights considerations.
How it works
The skill explains that Odyssey generates video one frame at a time in real time, conditioned on past frames and current user input, producing an interactive stream rather than a downloadable clip. It defines the user-in-the-loop model behavior, per-frame latency (e.g., ~40 ms/frame for Odyssey-1 up to 30 fps; ~50 ms/frame for Odyssey-2, ~20 fps; up to 24 fps for Starchild-1), and that every frame is generated rather than pre-authored. It contrasts Odyssey’s causal autoregressive generation with traditional offline clip models and details model lineup, access routes (web experiences at experience.odyssey.ml; API via developer.odyssey.ml), and the nascent nature of programmatic access. It also covers production suitability, coherence limits (roughly 1–2 minutes of reliable coherence, with drift over longer horizons), and lack of exportable assets by default. It lists per-user GPU costs (historical figure: $1–$2 per user-hour) and notes that interactive sessions consume live GPU resources rather than serving cached video. It provides decision heuristics to choose Odyssey when live steering is required and the deliverable tolerates dreamlike, non-persistent visuals, or experiential demos. It cautions that there are no published guarantees for AAA-style shipping, persistent worlds, or exportable assets, and emphasizes checking current API terms and quotas at developer.odyssey.ml. It also flags content/rights cautions, including open questions about ownership of generated content and IP concerns, and the need for consent/likeness diligence with real people depicted in generated streams.
When to use it
Use when a request involves live, user-steered video that reacts in the moment (keyboard, text, speech, or controller input), or when deciding between Odyssey interactive generation and offline generators or game-engine pipelines. Favor Odyssey for demos, prototypes, or experiential installations where dreamlike aesthetics and real-time interactivity are central and long-term deterministic persistence or downloadable assets are not required. Do not route here for offline, fixed clips, exportable assets, or long-form production pipelines that require guaranteed reproducibility.
What it can touch
Public web interfaces for Odyssey-2 (experience.odyssey.ml) and Agora-1 (agora.odyssey.ml), and the emerging Odyssey-2 Pro API surfaces (interactive streams, viewable/broadcast streams, Simulate). Authentication uses an API key via the API docs at documentation.api.odyssey.ml and dashboard at developer.odyssey.ml/dashboard. Endpoints, quotas, and terms are described as real but nascent; confirm current status at developer.odyssey.ml before promising integration.
Caveats
Maturity is research-grade and not production-ready for guaranteed reliability. Visual fidelity is unstable with possible blur and text issues; coherence is time-bounded; world persistence is weak with possible drift; no exportable assets by default; per-user GPU cost exists and scales with concurrent viewers; no published consumer price tiers; programmatic access is early with quotas/terms to confirm; content/rights questions remain about ownership and IP resemblance; consent/licensing considerations apply for real-person likeness in generated outputs.
# Odyssey interactive video (world models) ## What this skill is for Odyssey builds **world models**: systems that generate video *one frame at a time in real time*, conditioned on the history of frames plus whatever input a user is streaming right now. The output is not a finished clip you download — it is an interactive stream the user drives. This skill helps an agent scope, plan, and honestly
What does the odyssey-interactive-video skill do?
Plan and scope production work with Odyssey's real-time interactive video world models (odyssey.ml — Odyssey-1/Odyssey-2/Starchild-1/Agora-1). Use when a request involves generating video that a user steers live with keyboard, text, speech, or controller input (streamed frames that respond in the moment), or when someone must decide between Odyssey-style interactive generation and offline clip generators (Sora/Veo/Kling) or downloadable 3D / game engines. Covers what is verified versus announced, current access routes, latency and coherence limits, feasible-now versus speculative use cases,
How do I install it?
Run `npx skills add calesthio/generative-media-skills --skill odyssey-interactive-video --agent claude-code` — it drops the skill into your project so the agent can pick it up. Swap the --agent value for codex, cursor or copilot if you use one of those.
Where does this skill come from?
From calesthio/generative-media-skills, a repository with 112 stars. We read it straight from the repository tree rather than a submitted listing, so what you see here is what is actually published.
Is a popular skill a good skill?
Not necessarily. Stars measure attention, not adoption — a repository can trend for a week and be abandoned. That is why we show the weekly change from our own snapshots next to the total, instead of a single flattering number.