Everless321/dYm

What it solves

dYm is a desktop application designed for content creators and researchers who need to collect and organize short-form video content from Douyin. It solves the problem of manually watching and tagging videos by automating the process of downloading watermark-free videos and using AI to analyze their content.

How it works

The application uses a multimodal Vision LLM (via an OpenAI-compatible API) to understand video content. It extracts key frames from videos using FFmpeg at configurable intervals and sends these frames to the AI model. For image carousel posts, it sends the original images directly. The AI then generates structured metadata—including tags, categories, summaries, and scene descriptions—which is stored in a local SQLite database for easy searching and filtering.

Who it’s for

It is built for social media managers, content creators, and researchers who need to scale their video collection and content analysis workflows.

Highlights

• AI-Driven Analysis: Automatically generates tags, summaries, and content ratings for videos and image posts.

• Watermark-Free Downloads: Supports batch downloading of videos and recording of live streams in original quality.

• Customizable AI Pipeline: Allows users to define custom prompts, select different Vision API models, and configure concurrency and rate limits.

• Local Data Management: Stores all metadata and video information in a local SQLite database for full user control.

• Smart Filtering: Enables filtering of content based on AI-generated tags and categories.