AI Video Pipeline

A full-stack AI video generation system. Generate cinematic videos from any concept — auto-storyboarded by OpenAI, narrated by ElevenLabs, rendered by ComfyUI + AnimateDiff, delivered via Cloudflare R2, and notified by Telnyx SMS.

Architecture

• Dynamic Storyboards: Describe any concept → OpenAI generates a full storyboard automatically
• Static Storyboards: Pre-built brand storyboards (e.g. Helios Digital)
• Script Polish: OpenAI refines narration scripts for target audience
• Narration: ElevenLabs Text-to-Speech
• Rendering: ComfyUI API with AnimateDiff (DreamShaper 8 + v3 motion model)
• Scene Stitching: ffmpeg auto-concatenation of rendered scenes
• Branding: Dedicated animated logo sequences (Solar Ignition, Vault-Forged, Ledger Pulse)
• Delivery: Cloudflare R2 object storage upload
• Notification: Telnyx SMS completion alerts

Project Structure

.
├── branding/
│   └── logo_sequences.json         # Animated logo definitions
├── integrations/
│   ├── cloudflare_r2.py            # Cloudflare R2 upload client
│   ├── comfyui_client.py           # ComfyUI API (AnimateDiff + image gen)
│   ├── elevenlabs_client.py        # ElevenLabs TTS
│   ├── openai_client.py            # OpenAI script refinement
│   ├── storyboard_gen.py           # Dynamic storyboard generation
│   └── telnyx_client.py            # Telnyx SMS notifications
├── outputs/                        # Generated videos, stills, audio
├── storyboards/
│   └── helios_master.json          # 12-scene Helios master storyboard
├── .env                            # API keys (git-ignored)
├── .env.example
├── notify_completion.py            # CLI: send SMS notification
├── render_logo.py                  # CLI: render animated logos
├── upload_assets.py                # CLI: upload to R2
└── run_video_pipeline.py           # Master orchestration script

Setup

1. The .env file is configured with your API keys.

2. Install dependencies:

pip install openai requests boto3 python-dotenv websocket-client

3. Make sure ComfyUI is running:

cd c:\ComfyUI && python main.py

Usage

Generate any video from a concept (auto-storyboard):

python run_video_pipeline.py --concept "Blockchain Payment Rails Explainer" --scenes 8

Use the Helios master storyboard:

python run_video_pipeline.py --concept "Helios Digital Institutional Promo" --brand helios --scenes 12

Dry run to preview the plan:

python run_video_pipeline.py --concept "DeFi Yield Aggregator" --scenes 6 --dry-run

Generate stills instead of video:

python run_video_pipeline.py --concept "Product Shots" --scenes 6 --stills-only

Custom storyboard file:

python run_video_pipeline.py --concept "My Custom Video" --storyboard my_storyboard.json

Render animated logos:

python render_logo.py --sequence solar_ignition
python render_logo.py --sequence vault_forged
python render_logo.py --sequence ledger_pulse

Advanced options:

python run_video_pipeline.py --concept "..." \
  --scenes 10 --width 768 --height 432 --fps 8 --frames 16 \
  --steps 25 --cfg 7.0 --style "cinematic dark luxury" \
  --audience "retail" --duration 120 \
  --no-upload --no-sms --no-tts