grok-video-workflow - GitLite

KO | EN

gitlite — search

#python #typescript #llm #go #rust #deep-learning #docker #macos #java #windows #kubernetes #vue #pytorch #openai

grok-video-workflow

(★ 122)

Codex workflow for Grok Imagine Video API generation, download, and frame review

파일 탐색기

최종 버전 다운로드 (.zip)

- README.md
- .gitkeep
- worldcup-fancam.example.txt
- grok-video.mjs
- review-video.mjs
.env.example
.gitignore
LICENSE
package-lock.json
package.json
README.md
README.zh-CN.md

# 설치 가이드

1. 코드 내려받기

git clone https://github.com/Rion-Wu-tech/grok-video-workflow

깃허브에서 프로젝트 코드 전체를 내 컴퓨터로 내려받습니다.

cd grok-video-workflow

방금 내려받은 프로젝트 폴더 안으로 이동합니다.

2. Node.js

쉬움 추천

사전 준비물

Git GitHub에서 프로젝트 코드를 내려받으려면 필요합니다.
Node.js npm 명령어를 쓰려면 Node.js가 설치되어 있어야 합니다. LTS(장기지원) 버전을 추천합니다.

npm install

package.json에 명시된 라이브러리들을 내려받아 설치합니다.

npm run video -- --prompt "A cinematic AI creator editing videos at midnight, vertical social media style" --duration 5 --aspect-ratio 9:16 --resolution 480p

이 명령어를 터미널에 그대로 입력해 실행하세요.

npm run models

이 명령어를 터미널에 그대로 입력해 실행하세요.

npm run video:standard -- --prompt "A cinematic AI creator editing videos at midnight, vertical social media style" --duration 5 --aspect-ratio 9:16 --resolution 480p

이 명령어를 터미널에 그대로 입력해 실행하세요.

npm run video:latest -- --prompt "A cinematic AI creator editing videos at midnight, vertical social media style" --duration 5 --aspect-ratio 9:16 --resolution 720p

이 명령어를 터미널에 그대로 입력해 실행하세요.

✅ 명령어 실행 후 터미널에 나타나는 주소(보통 http://localhost:3000 형태)를 브라우저에서 열어보세요.

이 레포의 README에 적힌 실제 명령어를 그대로 가져왔습니다.

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

# Grok Video Workflow

**Languages:** English | [简体中文](./README.zh-CN.md)

A small local CLI workflow for generating videos with the xAI Grok Imagine Video API.

If this project helps you, please give it a ⭐ Star!

## What It Does

This project wraps Grok Imagine Video into a practical local workflow:

- text-to-video
- reference-to-video with local images or HTTPS image URLs
- async polling
- automatic video download
- metadata JSON output
- cost estimates
- contact sheet frame review
- Codex-friendly command usage

It is designed to work well with Codex: ask Codex to write or compress prompts, run the CLI, inspect the generated contact sheet, and decide whether to rerun.

## Important Notes

- This uses the xAI API, not the Grok web app quota.
- You need xAI API credits or billing enabled in the xAI Console.
- Do not commit your `.env` file.
- Grok prompts currently have a 4096 character limit.
- Reference-to-video supports up to 10 seconds.
- Video generation can produce unstable text, hands, or identity details. Review before publishing.

## Setup

```bash
git clone https://github.com/Rion-Wu-tech/grok-video-workflow.git
cd grok-video-workflow
npm install
cp .env.example .env
```

Edit `.env`:

```env
XAI_API_KEY=your-xai-api-key
XAI_VIDEO_MODEL=grok-imagine-video
```

## Generate A Text-To-Video Clip

```bash
npm run video -- --prompt "A cinematic AI creator editing videos at midnight, vertical social media style" --duration 5 --aspect-ratio 9:16 --resolution 480p
```

The workflow includes both xAI video models. You can list them locally:

```bash
npm run models
```

Use the standard model for lower-cost routine generations:

```bash
npm run video:standard -- --prompt "A cinematic AI creator editing videos at midnight, vertical social media style" --duration 5 --aspect-ratio 9:16 --resolution 480p
```

Use the newer preview model when you want to test the latest quality:

```bash
npm run video:latest -- --prompt "A cinematic AI creator editing videos at midnight, vertical social media style" --duration 5 --aspect-ratio 9:16 --resolution 720p
```

## Generate A Reference-To-Video Clip

Put your reference image in `examples/`, then run:

```bash
npm run video -- --prompt-file prompts/worldcup-fancam.example.txt --reference-image examples/your-storyboard.png --duration 10 --aspect-ratio 1:1 --resolution 720p --prefix worldcup-fancam
```

The CLI prints:

```text
request_id
status
video_url
saved_video
metadata
```

Generated files are saved to `outputs/` by default.

## Review The Result

Create a contact sheet:

```bash
npm run review -- --video outputs/your-video.mp4
```

Review checklist:

- Is the action sequence continuous?
- Did the subject's identity stay consistent?
- Are hands and fingers acceptable?
- Is text or scoreboard content stable?
- Did the scene jump unexpectedly?

## Cost Estimate

At the time this workflow was created, xAI's public pricing listed Grok Imagine Video around:

```text
grok-imagine-video:
  480p: $0.05 / second
  720p: $0.07 / second

grok-imagine-video-1.5-preview:
  480p: $0.08 / second
  720p: $0.14 / second
```

Examples:

```text
5s 480p  ~= $0.25
10s 480p ~= $0.50
10s 720p ~= $0.70
```

Always check the xAI Console and official pricing page before large batches.

## CLI Options

```text
--prompt <text>              Text-to-video prompt.
--prompt-file <path>         Read prompt from a UTF-8 text file.
--reference-image <path|url> Reference image. Repeat up to 7 times.
--duration <seconds>         1-15. Reference-to-video max is 10. Default: 5.
--aspect-ratio <ratio>       16:9, 9:16, 1:1, 4:3, 3:4, 3:2, 2:3. Default: 16:9.
--resolution <value>         480p or 720p. Default: 480p.
--output-dir <path>          Default: outputs.
--prefix <name>              Output filename prefix. Default: grok-video.
--poll-interval <seconds>    Default: 5.
--timeout-minutes <minutes>  Default: 20.
--request-id <id>            Poll and download an existing request.
--model <name>               Default: grok-imagine-video. Or set XAI_VIDEO_MODEL.
--list-models                Show supported video models.
--no-download                Print hosted URL without saving video.
```

## Codex Prompt Examples

```text
Use this repo to generate a 5 second 9:16 Grok video. First compress my prompt under 4096 characters, then run the CLI and review the contact sheet.
```

```text
Generate a reference-to-video clip from examples/storyboard.png, duration 10 seconds, 720p, then create a contact sheet and tell me whether it needs a rerun.
```

## Why Not A Codex Official Grok Plugin?

Codex can run local tools and can be extended through scripts, skills, and MCP servers. xAI Grok video generation is exposed through xAI's API. This repository wraps that API into a local workflow that Codex can operate.

## Author

Created by [Rion-Wu-tech](https://github.com/Rion-Wu-tech).
X/Twitter: [@rionaifantasy]

## Safety

Respect image rights, likeness rights, trademarks, and event broadcast rights. Do not present AI-generated event footage as real footage.

If this project helps you, please give it a ⭐ Star!