Rocketreels is an AI-powered video generation pipeline that automatically creates engaging short-form videos (like TikTok, YouTube Shorts, or Instagram Reels) from simple subject inputs. Give it a topic, and it does the rest - generating script, narration, visuals, editing, and subtitles.
- Complete Automation - From idea to finished video with one command
- AI-Generated Content - Uses GPT for scripts, DALL-E/Gemini for images, and AI voice synthesis
- Professional Editing - Automatic Ken Burns effects, properly timed transitions, and synchronized subtitles
- Customizable - Configure resolutions, aspect ratios, fonts, voices, and more
- Batch Processing - Generate multiple videos on a subject in parallel
See it in use:
And the result of that process (only a few minutes)
output.mp4
- Content Generation - Creates script/transcript from your topic using GPT
- Audio Synthesis - Converts script to natural-sounding narration using AI voices
- Image Generation - Produces relevant images for each section using DALL-E or Gemini
- Video Assembly - Applies Ken Burns effects to images, synchronizes with audio
- Subtitle Integration - Adds perfectly timed subtitles with customizable styling
- Python 3.12
- FFmpeg
- Pillow/PIL
- libsm6 (for OpenCV support)
- API keys for:
- OpenAI (required)
- Google Gemini (optional)
- ElevenLabs (optional)
# Clone the repository
git clone https://github.com/romanobro56/rocketreels.git
cd rocketreels
# Install dependencies with Poetry
poetry install
# Create config.py with your API keys
cat > config.py << EOF
OPENAI_API_KEY = "your-openai-key"
GEMINI_API_KEY = "your-gemini-key" # Optional
ELEVENLABS_API_KEY = "your-elevenlabs-key" # Optional
EOF
The easiest way to get started is with the interactive script:
poetry run python interactive_create_video.py
This will guide you through all the options for generating a video.
For more control, use the command-line script:
poetry run python generate_video.py --subject "History of Pirates" --idea-seed 3 --orientation vertical --quality HD --voice onyx
# Content Generation
--subject TEXT Subject of the video [required]
--idea-seed INTEGER Idea seed (0-9) [default: 0]
--batch INTEGER Number of videos to generate in batch [default: 0]
--image-generator TEXT Image generator to use (openai/gemini) [default: openai]
--tts-provider TEXT TTS provider to use (openai/elevenlabs) [default: openai]
--voice TEXT Voice to use for TTS [default: onyx]
# Video Editing
--orientation TEXT Video orientation (vertical/horizontal/square) [default: vertical]
--quality TEXT Video quality (HD/4K/8K) [default: HD]
--frame-rate INTEGER Video frame rate [default: 30]
--font-size TEXT Font size for subtitles [default: extra_large]
Generate multiple videos on a topic in one go:
poetry run python generate_video.py --subject "Space Exploration" --batch 5
content_generation_workflow.py
- Main content generation pipelinevideo_editing_workflow.py
- Video editing and assembly pipelinevideo_pipeline_workflow.py
- Combines generation and editing processesgeneration/
- Script, audio, and image generation modulesediting/
- Video and subtitle processing modulesmodels/
- Data models and configurationsutils/
- Helper utilitiesediting_config/
- Video editing configurations
Video quality and style settings can be found in editing_config/editing_config.py
.
The system supports various aspect ratios and quality levels:
- Orientations: vertical (9:16), horizontal (16:9), and square (1:1)
- Quality: HD, 4K, and 8K
- Visual styles: dynamic, dark, light, default
This project uses several AI services:
- OpenAI (GPT, Whisper, DALL-E)
- Google Gemini (for image generation)
- ElevenLabs (for voice synthesis)
Don't you dare try and contribute!
Made with hate in my heart by Roman