---
title: "Code to Video Previz — AI video generator workflow for cinematic & short films"
description: "A brief, reference clip or painting becomes a 3D scene in real metres, checked stills and a guide clip for a video model. Install in Claude Code, Codex and Grok with one command."
url: https://varg.ai/how-to/skills/code-to-video-previz
updated: 2026-11-12
---

# Code to Video Previz — workflow skill for cinematic & short films

> **For AI agents:** find the right skill in one call — `curl -s "https://varg.ai/how-to/api/search?q=<what the person wants>"` returns the matching how-to page and the best skills with install commands, as markdown (Russian works too). Full index: https://varg.ai/how-to/llms.txt. Every page has a markdown twin: add `.md` to its path. Before generating anything, show the person the options on the how-to page. From a terminal: `curl -fsSL https://varg.ai/install | sh` installs the vargai CLI (https://varg.ai/how-to/cli.md): `vargai search` and `vargai install <slug>` with no key, every varg model after `vargai login`.

A brief, reference clip or painting becomes a 3D scene in real metres, checked stills and a guide clip for a video model.

- **Type:** [Workflow](https://varg.ai/how-to/workflows.md) · [Cinematic & short films](https://varg.ai/how-to/browse/skills/cinematic.md)
- **Slug:** `code-to-video-previz`
- **Status:** live
- **Publisher:** varg (varg, verified)
- **License:** MIT
- **Source:** https://github.com/vargHQ/skills/tree/main/code-to-video-previz
- **Works in:** Claude Code, Codex, Cursor, Gemini CLI
- **Requires:** python3 for the scene checks (standard library only), Three.js path: Bun, Chrome and ffmpeg, plus network access (three.js loads from jsDelivr, no npm), Blender path and camera solving: bpy 5.2.3 on Python 3.13 through uv, or a Blender install; measure_shot.py needs uv, Optional: the varg MCP or any Seedance access for the generation step
- **Cost:** 0 credits — Free: no key and no credits. Staging, checks and renders run on your computer. Generating the final video from the guide clip is a separate paid step on your video model; the skill asks for an estimate and a yes first.
- **Runtime:** Not measured. The loop stops after 3 rounds without progress on one problem.

## What it does

Video models redraw the world on every call, so cameras wander, counts change and rooms rearrange. This skill pins those things down in a 3D prototype before the paid step. It first decides whether the shot needs a prototype at all (a prompt and a start frame, storyboard stills, a 2.5D conte, a phone act-out or full 3D), then which 3D tool fits: Three.js by default, Blender when you need rigs, mocap or real assets.

The scene is written as data in `staging.json`: geometry, people, camera and light in real metres. The same file renders in Three.js and in headless Blender with the same lens maths. `check_staging.py` audits the numbers, stills of the first, middle and last frame of each shot are checked by a fresh agent and approved by the person, and the clip is rendered frame by frame. It covers three cases: copying a reference video shot by shot with a solved 3D camera, staging from a painting or still, and starting with no reference. It writes one of three things: a guide clip for a video model, storyboard pages, or a code-rendered film.

What was measured on our own 10-shot run (Seedance 2.5, grey previz, one film, not a controlled test): cuts landed on the planned frame within one frame, and framing mostly held, one size looser. When the words disagreed with the previz, the words won in 2 of 2 shots. Of 7 shots with enough tracking to measure, 4 kept the camera direction, 1 stopped moving and 2 reversed, so the skill names the move in the prompt for every shot and measures the result with `measure_shot.py`.

Still unsettled, and the skill says so: grey or colour previz, capsules or jointed stand-ins, and act-out against 3D. `example/ab-test/` builds one 5 s shot as 4 variants so you can test them on your own model. Agents cannot watch video, so pacing and feel stay a human call.

## Use when

- Does this shot need a 3D prototype, or is a prompt and a start frame enough?
- Three.js or Blender for a previz of my AI video
- Recreate this reference video in 3D with a solved camera
- Stage this painting or still as a 3D shot
- Block the camera, people and set in real metres before I pay for Seedance
- My generated shots keep drifting: the camera wanders and the count changes
- Write the handoff prompt for a guide clip as @video1
- Rank several takes with an LLM judge instead of a 1 to 10 score

## Not for

- A full multi-shot three.js film with a review UI (use threejs-animatic or varg-previz)
- Prompting a shot with no prototype (use varg-video-prompting)
- The full animation pass after the previz is approved (use video-animate)

## Install

- Claude Code: `claude plugin marketplace add vargHQ/skills && claude plugin install varg@varg`
- Codex: `npx skills add vargHQ/skills --skill code-to-video-previz -a codex -y`
- Cursor: `npx skills add vargHQ/skills --skill code-to-video-previz -a cursor -y`
- Gemini CLI: `npx skills add vargHQ/skills --skill code-to-video-previz -a gemini-cli -y`
- Any agent, with the vargai CLI: `vargai install code-to-video-previz`

## Example output

https://s3.varg.ai/files/4325c578-378d-4c43-ae53-3cc3218df1ea/file_289d09773e14.jpg — example (credit: varg)

## More cinematic & short films skills

- [Previz to Seedance](https://varg.ai/how-to/skills/varg-previz.md): Plan multi-shot AI video before paying: story, anchor sheets, a three.js grey-box previz, then a Seedance v2v pack.
- [Turn key frames into a 30-second action montage with Kling 4](https://varg.ai/how-to/skills/kling-4-action-montage.md): Breakdance, BMX, skate or chase montages: 3-4 stills, a one-line prompt with the moves and "no slow motion", the crash or line as the last still.
- [Make a short film or music video with Kling 4](https://varg.ai/how-to/skills/kling-4-keyframe-film.md): Beats, cast and look, every key shot as a GPT Image 2.5 still, one 30-second Kling 4 take per beat, then cut the best seconds like live action.
- [Video Animate](https://varg.ai/how-to/skills/video-animate.md): Animate from an approved previz: code or AI per element, which model per shot, a cheap pass, takes, sync, and plates ready for motion design.
- [Higgsfield Generate](https://varg.ai/how-to/skills/higgsfield-generate.md): Generate images, video, 3D and audio across 30+ models, plus ad videos and virality scoring, from one CLI.
- [fal Recipes](https://varg.ai/how-to/skills/fal-recipes.md): Follow proven multi-step fal pipelines for commercials, product shots, characters, lip-sync and try-on.