MCPcopy Index your code
hub / github.com/RafaelGodoyEbert/ViralCutter

github.com/RafaelGodoyEbert/ViralCutter @main

Chat with this repo
repository ↗ · DeepWiki ↗ · + Follow
124 symbols 440 edges 32 files 41 documented · 33%
What it actually does AI analysis from the code graph — generated when you open this
loading…
README

ViralCutter

Discord

100% Free, Local, and Unlimited Open-Source Alternative to Opus Clip
Turn long YouTube videos into viral shorts optimized for TikTok, Instagram Reels, and YouTube Shorts – with state-of-the-art AI, dynamic captions, precise face tracking, and automatic translation. All running on your machine.

Stars Forks Open in Colab

EnglishPortuguês

Why is ViralCutter a "Game Changer"?

Forget expensive subscriptions and minute limits. ViralCutter offers unlimited power on your own hardware.

Feature ViralCutter (Open-Source) Opus Clip / Klap / Munch (SaaS)
Price Free & Unlimited $20–$100/mo + minute limits
Privacy 100% Local (Your data never leaves your PC) Upload to third-party cloud
AI & LLM Flexible: Gemini (Free), GPT-4, Local GGUF (Offline) Only what they offer
Face Tracking Split Screen (2 faces), Active Speaker (Exp.), Auto Basic or extra cost
Translation Yes (Translate captions to 10+ languages) Limited features
Editing Export XML to Premiere Pro (Beta) Limited web editor
Watermark ZERO Yes (on free plans)

Professional results, total privacy, and zero cost.

Key Features 🚀

  • 🤖 AI Viral Cut: Automatically identifies hooks and engaging moments using Gemini, GPT-4, or Local LLMs (Llama 3, DeepSeek, etc).
  • 🗣️ Ultra-Precise Transcription: Powered by WhisperX with GPU acceleration for perfect subtitles.
  • 🎨 Dynamic Captions: "Hormozi" style with word-by-word highlights, vibrant colors, emojis, and full customization.
  • 🎥 Auto Camera Direction:
    • Auto-Crop 9:16: Transforms horizontal to vertical while keeping the focus.
    • Smart Split Screen: Detects 2 people talking and automatically splits the screen.
    • Active Speaker (Experimental): The camera cuts to whoever is speaking.
  • 🌍 Video Translation: Automatically generate translated subtitles (e.g., English Video -> Portuguese Subtitles).
  • 💾 Quality & Control: Choose resolution (up to 4K/Best), format output, and save processing configurations.
  • Performance: Transcription with "slicing" (process 1x, cut N times) and ultra-fast installation via uv.
  • 🖥️ Modern Interface: Gradio WebUI, Dark Mode, Project Gallery, and integrated Subtitle Editor.

Web Interface (Inspired by Opus Clip)

WebUI Home Intuitive control panel with fine-tuning for AI and rendering.

WebUi Library Library: OpusClip-style gallery and intuitive controls

Local Installation (Super Fast ⚡)

Prerequisites (From Scratch Setup)

To run ViralCutter on a fresh computer, you need to install the following core tools:

  1. Visual Studio C++ Build Tools Required to compile insightface and avoid "Cpp/Visual Studio" setup errors.
  2. Download Microsoft C++ Build Tools.
  3. Run the installer and check the "Desktop development with C++" box.
  4. Ensure Windows 10/11 SDK and MSVC v143 - VS 2022 C++ are checked on the right panel, then click install. Restart your PC if prompted.

  5. Python (3.10.x or 3.11.x recommended)

  6. Download from python.org/downloads.
  7. ⚠️ VERY IMPORTANT: On the very first setup screen, mark the checkbox "Add Python to PATH" at the bottom before clicking install.

  8. FFmpeg (Audio/Video Processing Engine)

  9. The easiest way on Windows is to open your terminal (PowerShell) as Administrator and run: winget install ffmpeg
  10. Restart the terminal and type ffmpeg -version to confirm it works.

  11. Video Card Drivers (NVIDIA)

  12. Keep your drivers updated (via GeForce Experience or the Nvidia website) to support CUDA 12.4+ acceleration.
  13. NVIDIA GPU is highly recommended for speed and local AI operations.

Step-by-Step Installation

  1. Install Dependencies via Script Open the ViralCutter folder and double-click one of the installers below:

    • install_dependencies.bat: Standard installation (Recommended). Faster and fail-proof. Uses cloud AIs like Gemini (Free) and GPT-4.
    • install_dependencies_advanced_LocalLLM.bat: Advanced installation. Dedicated for users who want to run full offline AIs on their hardware (Llama 3, etc). Requires a good GPU and C++ Build Tools.

    (Both use the uv package manager to set everything up automatically).

  2. Configure AI (Optional)

    • Gemini (Recommended/Free): Add your key in api_config.json.
    • Local (GGUF): Download your favorite .gguf models and place them in the models/ folder. ViralCutter will detect them automatically.
  3. Run

    • Double-click run_webui.bat to open the interface in your browser.
    • Or use python main_improved.py for the CLI version.

Output Examples

Viral Clip with Highlight Captions

Direct Comparison: Opus Clip vs ViralCutter (same input video)

2-Face Split Screen Mode

Roadmap (TODO)

  • [x] Release code
  • [ ] Permanent Demo on Hugging Face Spaces
  • [x] Two face in the cut (Split Screen)
  • [x] Custom caption and burn
  • [x] Make the code faster
  • [x] 100% Local AI Models (Ollama/Llama/GGUF)
  • [x] Automatic caption translation
  • [x] The cut follows the face as it moves
  • [x] XML Export to Premiere Pro (Beta)
  • [ ] Automatic background music (Auto-Duck)
  • [ ] Direct upload to TikTok/YouTube/Instagram
  • [ ] More framing formats (beyond 9:16)
  • [ ] Optional Watermark

Contribute!

ViralCutter is community-maintained. Join us to democratize AI content creation! - Discord: AI Hub Brasil - Github: Give us a ⭐ star if this project helped you!

Current Version: 0.8v Alpha ViralCutter: Because viral clips shouldn't cost a fortune. 🚀

Core symbols most depended-on inside this repo

convert_color_to_ass
called by 12
webui/app.py
substituir_texto
called by 6
scripts/translate_json.py
finalize_video
called by 6
scripts/edit_video.py
resize_with_padding
called by 5
scripts/one_face.py
crop_center_zoom
called by 5
scripts/one_face.py
get_uid
called by 5
scripts/export_xml_lib/xml_generator.py
timestamp_to_srt
called by 4
scripts/export_xml_lib/utils.py
get_mode_avg
called by 4
scripts/export_xml_lib/xml_generator.py

Shape

Function 119
Method 3
Class 1
Route 1

Languages

Python100%

Modules by API surface

webui/app.py18 symbols
scripts/edit_video.py12 symbols
scripts/translate_json.py10 symbols
scripts/create_viral_segments.py8 symbols
scripts/transcribe_video.py7 symbols
webui/subtitle_editor.py6 symbols
webui/subtitle_handler.py5 symbols
webui/library.py5 symbols
scripts/export_xml_lib/xml_generator.py5 symbols
i18n/i18n.py5 symbols
scripts/one_face.py4 symbols
scripts/face_detection_insightface.py4 symbols

For agents

$ claude mcp add ViralCutter \
  -- python -m otcore.mcp_server <graph>

⬇ download graph artifact

Ask about this repo answers extend the page