Previz to Seedance — workflow skill for cinematic & short films
For AI agents: find the right skill in one call —
curl -s "https://skills.varg.ai/api/search?q=<what the person wants>"returns the matching use case and the best skills with install commands, as markdown (Russian works too). Full index: https://skills.varg.ai/llms.txt. Every page has a markdown twin: add.mdto its path. Before generating anything, show the person the options on the use-case page.
Plan multi-shot AI video before paying: story, anchor sheets, a three.js grey-box previz, then a Seedance v2v pack.
- Type: Workflow · Cinematic & short films
- Slug:
varg-previz - Status: beta
- Publisher: varg (varg, verified)
- License: MIT
- Source: https://github.com/vargHQ/skills/tree/main/varg-previz
- Works in: Claude Code
- Requires: python3 and a browser (previz app), bun + playwright-core + Chrome or Chromium (clip capture), ffmpeg (pack build), varg MCP connection or VARG_API_KEY (generation)
- Cost: ~65–180 credits / run (480p, as billed) — Story, storyboard, scene grouping and the previz are free. Seedance 2.5 v2v runs of 4–11 s at 480p billed about 16 credits per second while estimate_cost quoted 315–361, so the skill quotes the estimate as a ceiling and reports the billed cost after.
What it does
Pre-visualises any multi-shot video (music videos, ads, UGC spots, trailers, explainers) so camera, geography, scale and timing are fixed for free before the paid step. The pipeline runs story structure → storyboard.json → character, product and location anchors → master scenes of 4–11 s → a three.js grey-box previz with auto cameras from plain storyboard words → a Seedance 2.5 video-to-video pack where each run gets its previz clip as @video1 plus refs for only what is in frame. build_pack.py --check blocks costly mistakes (runs without image refs, refs under 300 px, runs over 15 s, shots in no run) before anything is spent. Built on a 36-shot music video and tested on a 6-shot product ad; in review as vargHQ/skills PR #24 and goes live when it merges.
Use when
- A previz, 3D animatic, camera blocking or camera approval pass
- Shot lists, scene grouping, or story structure for a clip or ad
- Character, product or vehicle sheets
- A Seedance 2.5 video-to-video pack
Not for
- 1–3 shot pieces with a simple camera: skip the previz and go straight to anchors and image-to-video
Install
- Claude Code:
claude plugin marketplace add vargHQ/skills && claude plugin install varg@varg - Codex:
npx skills add vargHQ/skills --skill varg-previz -a codex -y - Cursor:
npx skills add vargHQ/skills --skill varg-previz -a cursor -y - Gemini CLI:
npx skills add vargHQ/skills --skill varg-previz -a gemini-cli -y - OpenClaw:
npx clawhub@latest install varg-previz
Models it drives: seedance_2_5, gpt_image_2_5, sync_v2_pro
Example output
https://skills.varg.ai/previews/varg-previz.mp4 — The skill's own free step, run for real: its three.js grey-box previz app (vargHQ/skills feature/varg-previz, commit 30620f7) playing the bundled 6-shot sample storyboard 'Morning Drop' (9:16, 12 s, 3 locations). Auto cameras come from the storyboard words (CU push-in, OTS, WS truck right, MS pull out); the set map shows each shot's camera frustums. Captured in headless Chrome at 1440x900, 24 fps, by seeking the timeline frame by frame. No generation credits. (credit: varg (varg-previz app, rendered locally), https://github.com/vargHQ/skills/pull/24)
Used for
- AI short film or series: A multi-shot story with consistent characters: script, storyboard, previz, shots, edit.
More cinematic & short films skills
- Higgsfield Generate: Generate images, video, 3D and audio across 30+ models, plus ad videos and virality scoring, from one CLI.
- fal Recipes: Follow proven multi-step fal pipelines for commercials, product shots, characters, lip-sync and try-on.
- fal Fan Cam: Put someone into a live sports broadcast: a realistic fan-cam cutaway with scoreboard and channel overlays.
- Gemini Omni Flash: Generate and edit video with Gemini Omni 1.1 Flash: text-to-video, first/last frames, extensions, edits.
- Melies: Generate images, video, movie posters and thumbnails from the terminal with 148 AI actors and style flags.
- AI Video Generation: Generate video with 40+ models from one CLI: Veo 3.1, Seedance 2.0, HappyHorse, Wan, Grok and more.