🎬 Autonomous AI Video Production Pipeline

Features • Installation Guide • Pipeline • Dashboard • Cost Optimization • Scheduling

🤖 Automate AI-powered stickman animation production, from script to YouTube

# Using Homebrew (recommended)
brew install python@3.11

# Verify
python3 --version
pip3 --version

sudo apt update
sudo apt install python3 python3-pip python3-venv -y
python3 --version

FFmpeg is required for video assembly and the slideshow effect.

brew install ffmpeg
ffmpeg -version

sudo apt install ffmpeg -y
ffmpeg -version

git clone https://github.com/saiedpod-bot/Stickman-Studio.git
cd Stickman-Studio

# Create virtual environment
python -m venv .venv

# Activate it:
# Windows:
.venv\Scripts\activate
# macOS / Linux:
source .venv/bin/activate

# Upgrade pip
pip install --upgrade pip

# Install all dependencies
pip install -r requirements.txt
pip install edge-tts          # Free local TTS (neural voices)

You need a Google Cloud account to use Gemini (AI script) and Imagen (AI images). The $300 free trial gives you 90 days of free credits — enough to produce thousands of videos.

1. Go to cloud.google.com/free
2. Click "Get started for free"
3. Sign in with your Google account (Gmail)
4. Fill in:
   • Country
   • Name & address (billing info — your card will NOT be charged, used only for verification)
   • Credit/Debit card (Google does a temporary $1 hold and refunds it)
5. ✅ You now have $300 in free credits + 90-day trial

⚠️ No charges without your consent. The free tier also includes many always-free products. You can set budgets and alerts in the console.

1. Go to console.cloud.google.com
2. At the top bar, click the project dropdown → New Project
3. Enter project name (e.g. stickman-studio)
4. Note the Project ID (e.g. stickman-studio-123456) — you'll need it later
5. Click Create

Enable these APIs for your project:

To enable each:

1. Click the Enable link above
2. Make sure your project is selected (top bar)
3. Click Enable

This is how the project authenticates with Google Cloud:

1. Go to Service Accounts
2. Click + Create Service Account
3. Name: stickman-studio-sa
4. Click Create and Continue
5. Under Grant access → Add roles:
   • Vertex AI User (roles/aiplatform.user)
   • Storage Object Admin (roles/storage.objectAdmin)
6. Click Done

1. In the service accounts list, click on the email of your new account
2. Go to Keys tab → Add Key → Create New Key
3. Choose JSON → Create
4. A .json file will download automatically — keep it safe!
5. Rename it if you like (e.g. stickman-studio-key.json)

1. Move the service account JSON key to the project root folder (Stickman-Studio/)

2. Copy the example env file:

   cp .env.example .env

3. Open .env in any text editor and fill in:

   # Your GCP project ID (from Step 7)
   GCP_PROJECT_ID=stickman-studio-123456
   
   # Path to the service account key (from Step 9)
   GOOGLE_APPLICATION_CREDENTIALS=stickman-studio-key.json
   
   # GCP region (keep default)
   GCP_LOCATION=us-central1

4. Save the file

# Download installer
curl -O https://dl.google.com/dl/cloudsdk/channels/rapid/GoogleCloudSDKInstaller.exe
# Run the installer (follow GUI prompts)

# After installation, authenticate:
gcloud auth application-default login

# Install
curl https://sdk.cloud.google.com | bash
exec -l $SHELL

# Authenticate
gcloud auth application-default login

# Make sure virtual environment is activated
# Windows: .venv\Scripts\activate
# macOS/Linux: source .venv/bin/activate

# Generate a video with default slideshow mode:
python orchestrator.py "How black holes work" --video-mode slideshow

# Generate with Veo animation (premium, costs credits):
python orchestrator.py "Why the sky is blue" --video-mode animation

What to expect:

[1/4] ✍️ Script → Gemini generates storyboard (5-10 sec)
[2/4] 🎨 Images → Imagen generates scene images (30-60 sec)
[3/4] 🎬 Video → ffmpeg Ken Burns zoom (10-20 sec)
[3.5] 🗣️ TTS → edge-tts generates narration (10-30 sec)
[4/4] 🎞️ Assembly → ffmpeg combines everything (10-20 sec)
✅ Video saved to: projects/how_black_holes_work/final.mp4

streamlit run app.py

Opens in your browser at http://localhost:8501

To enable automatic publishing to YouTube:

1. Go to Google Cloud Console → Credentials
2. Click + Create Credentials → OAuth Client ID
3. Application type: Desktop app
4. Name: Stickman Studio YouTube Uploader
5. Click Create
6. Click Download JSON — rename it to client_secrets.json
7. Move client_secrets.json to the project root folder
8. First upload will open your browser for OAuth consent:
   • Sign in with your YouTube channel's Google account
   • Click Advanced → Go to App (unsafe) → Allow
9. Token is cached in youtube_token.json for future runs
10. ✅ Done! All future uploads will be automatic.

Topic Input
    │
    ▼
┌─────────────────────────────────────────────────────┐
│  Phase 1: Script (Gemini 2.5 Flash)                 │
│  → Viral hook + scene-by-scene storyboard.json      │
└─────────────────────────────────────────────────────┘
    │
    ▼
┌─────────────────────────────────────────────────────┐
│  Phase 2: Images (Imagen 3.0)                       │
│  → Character reference + consistent scene images    │
└─────────────────────────────────────────────────────┘
    │
    ▼
┌─────────────────────────────────────────────────────┐
│  Phase 3: Video                                     │
│  ┌─ slideshow: ffmpeg Ken Burns zoom (FREE) ──────┐│
│  └─ animation:  Veo 2.0 image-to-video (premium)  ┘│
└─────────────────────────────────────────────────────┘
    │
    ▼
┌─────────────────────────────────────────────────────┐
│  Phase 3.5: Narration (edge-tts — FREE)             │
│  → Per-scene MP3 with neural voices                 │
└─────────────────────────────────────────────────────┘
    │
    ▼
┌─────────────────────────────────────────────────────┐
│  Phase 4: Assembly (ffmpeg — local)                 │
│  → Concatenate + overlay audio + BGM → final.mp4   │
└─────────────────────────────────────────────────────┘
    │
    ▼
┌─────────────────────────────────────────────────────┐
│  Phase 5: YouTube Upload (OAuth 2.0)                │
│  → Auto-publish as public with title, desc, tags    │
└─────────────────────────────────────────────────────┘

Pre-generated example videos showing the pipeline output (also available on the Releases page):

All samples are in samples/ directory. Generate your own with:

python orchestrator.py "Your topic here" --video-mode slideshow

Launch the full GUI:

streamlit run app.py

Tabs:

• Live logs — real-time pipeline output in st.status
• Progress bars — per-phase, per-video tracking
• System Health — 🟢 Green / 🟡 Yellow / 🔴 Red indicator
• Kill Switch — immediately halt all autonomous processes
• Estimated API Usage — remaining Gemini/Imagen calls
• System Logs — expandable panel with last 100 log lines

Default mode is slideshow to minimize costs.

# Via dashboard: Schedule tab → Toggle On

Randomised intervals avoid YouTube pattern detection.

# Via dashboard: Monthly tab → Start Monthly Plan

• 1–3 videos/day, randomly assigned
• 5-day blocks with rotating publishing windows (08:00, 10:00, 14:00, 16:00, 20:00)
• Persists to system_state.json — survives server reboots
• Kill switch available in the UI

# CLI: process all ideas in daily_plan.json
python -c "from scheduler import start_batch_production; start_batch_production()"

# Full autonomous cycle (plan → produce → upload)
python -c "from scheduler import run_autonomous_cycle; run_autonomous_cycle('Science')"

stickman_studio/
├── app.py                    # Streamlit dashboard
├── orchestrator.py           # CLI + importable pipeline runner
├── content_planner.py        # Gemini → viral video ideas
├── scheduler.py              # Batch + autonomous + monthly scheduler
├── uploader.py               # YouTube OAuth 2.0 upload + auto-publish
├── tts_engine.py             # edge-tts (free local TTS)
├── storage.py                # GCS upload/download
├── ai_engine.py              # Module-level wrappers for all phases
├── requirements.txt
├── .env.example → .env       # Configuration
├── client_secrets.json       # YouTube OAuth (user-provided)
├── samples/                   # Pre-generated example videos
├── assets/
│   └── bgm.mp3               # Optional background music
├── stickman_studio/
│   ├── config.py             # .env loading + Vertex AI init
│   ├── logging_setup.py      # Console + file logging
│   ├── models.py             # Scene / StoryBoard dataclasses
│   ├── retry.py              # Tenacity retry decorator
│   └── phases/
│       ├── phase1_script.py      # Gemini storyboard
│       ├── phase2_images.py      # Imagen 3.0 images
│       ├── phase3_video.py       # Veo 2.0 video
│       ├── phase3_slideshow.py   # Ken Burns zoom
│       ├── phase4_assembly.py    # ffmpeg concat + audio
│       └── phase4_subtitles.py   # MoviePy subtitles

stickman-studio  ai-video-generation  google-vertex-ai
gemini  imagen  veo  youtube-automation  content-creator
python  streamlit  edge-tts  text-to-video
free-tts  ai-animation  video-pipeline

Add these to your GitHub repo → Settings → Topics for discoverability.

MIT © 2026 — Free to use, modify, and distribute.

Made with ❤️ and 🤖
_{Automate your content. Own your audience.}