Invite-only testers. 1–3 short videos/month when approved, 5–10 scene/build tries, watermarked exports, 7-day retention.
Cost risk: high if opened broadly.Public site: gallery, docs, pricing, and request access. Private dashboard: authenticated real generation with quotas, queue, status polling, manifests, and artifact downloads. Controlled beta — not unlimited public rendering.
MCStory keeps the generator real: approved users can submit jobs to the authenticated dashboard. The public landing page does not expose unauthenticated render endpoints.
Marketing, gallery, examples, pricing, API docs, and contact. No anonymous generation, no raw worker logs, no local paths.
Login/session required, scoped API token, queue, rate limits, run_manifest.json, artifact storage, and status polling for film / scene / model jobs.
Admin-only queue monitor, worker health, retries, kill switch, abuse review, quota ledger, and cost controls before broader rollout.
Current prototype: the local webapp can call real generation APIs when run in an approved environment. Public deployment should first add auth backend, durable queue, quotas, and signed artifact delivery.
This panel represents the private beta generator. In production, every button below is protected by login, CSRF/session checks, API tokens, quota, queue, and run ownership.
Every one generated from a single sentence · scene / cast / dialogue / camera, all automatic
One sentence in, a fully-directed short film out — six automated steps
The same text→3D engine builds any structure from one sentence
Type one sentence — watch the 3D scene build itself, block by block




Proposed ranges while costs are measured. No public payments are enabled from this prototype.
Invite-only testers. 1–3 short videos/month when approved, 5–10 scene/build tries, watermarked exports, 7-day retention.
Cost risk: high if opened broadly.Individual creators. ~5–10 short videos or 30–50 scene/model jobs, 1 concurrent job, MP4/SRT/.schem exports.
Cost risk: medium; cap retries.Serious creators. ~20–50 videos or 150–300 scene/model jobs, script editing, 9:16 + 16:9, manifests, priority queue.
Cost risk: medium-high.Shared workspace, pooled quota, project folders, review links, reusable presets, webhooks, 5–10 concurrent jobs.
Cost risk: high; capacity planning needed.Dedicated workers, SSO, private storage, custom packs, audit logs, private deployment, contracted volume and SLA.
Cost risk: very high; scope tightly.Scoped API keys, REST jobs, webhooks, signed artifact URLs. Scene/model API first; video API by approval.
Cost risk: high; enforce spend caps.MCStory API should start with text → scene/model/.schem because it is cheaper and safer than full video. Full film generation remains approval-gated until worker capacity and abuse controls are proven.
POST /api/v1/runs
Authorization: Bearer mcs_live_***
{
"mode": "scene",
"prompt": "a red five-story Chinese pagoda with a gold roof",
"outputs": ["schem", "preview_png", "manifest"],
"webhook_url": "https://example.com/mcstory-webhook"
}
→ { "run_id": "run_20260605_abc123", "status": "queued", "status_url": "/api/v1/runs/run_20260605_abc123" }
For controlled beta access, API partnerships, Studio plans, or private cloud discussions, email:
huoz@berkeley.edu