AI-Rider Video | AI Video Workbench

中文 | English

AI-Rider Video is a desktop-first AI video workbench evolving from the original "first frame + last frame + prompt" workflow into a template-driven, multi-model creation product.

Current stable version: 2.3.1

Documentation

• Version History
• 2.0.0 PRD Draft
• 1.x Release Archive
• 1.1.3 Release Notes
• Desktop Environment Setup
• Desktop App Roadmap
• Desktop Build Automation

2.x Overview

The 2.x line is the first real product-system upgrade for AI-Rider Video.

• 2.0.x: model adaptation, provider abstraction, activation-code groundwork
• 2.1.x: product-expression and workbench UX upgrades
• 2.2.x: workspace expansion, reusable assets, and model/template entry pages
• 2.3.x: commercialization and ecosystem integration

The current 2.3.1 build follows the 2.3.0 licensing rollout by completing the desktop packaging and release automation path:

• keeps the 2.3.0 activation and licensing flow
• adds Windows packaging configuration and icon assets
• adds GitHub Actions automation for mac + win builds
• fixes CI dependency installation conflicts for build runners

Features

• Template-driven video generation workflows
• First + last frame generation
• First-frame generation
• Text-to-video generation
• Experimental video extension entry
• Multi-model support with a registry-based capability layer
• Local cost estimation based on token usage
• Advanced panels for API endpoints, JSON passthrough, and debugging
• Backend job creation, polling, result inspection, and local downloads
• First-launch activation onboarding
• License validation and remaining-use control
• License generation, freeze, revoke, and device reset tooling

Start

Install desktop dependencies

cd '/Users/silence/Documents/AI-Rider｜首尾帧视频输出桌面客户端'
npm run setup:desktop
npm run check:env

Required local tools:

• Node.js 20 LTS or newer
• npm 10 or newer
• Git
• ffmpeg / ffprobe

Install media tools on macOS with Homebrew:

brew install ffmpeg

Start the desktop app

npm run start:desktop

Start the web app

cd '/Users/silence/Documents/AI-Rider｜首尾帧视频输出桌面客户端'
npm run start:web

Open:

http://127.0.0.1:3100

Default directories

• UI files: public/
• Downloaded videos: downloads/

Notes

Default endpoints are prefilled as:

• Create: https://ark.cn-beijing.volces.com/api/v3/contents/generations/tasks
• Query: https://ark.cn-beijing.volces.com/api/v3/contents/generations/tasks/{task_id}

If your account uses different field names or endpoint templates, update them in the UI without changing code.

Example for extra JSON parameters:

{
  "return_last_frame": true
}

The current implementation sends the first frame and the last frame as two reference images. Depending on the exact Volcengine capability enabled on your account, the last frame may work as a hard ending constraint or as an additional reference image.