Sora Streamlit Studio

An end-to-end Streamlit app for creating watermark-free videos with OpenAI’s Sora API — featuring reference images, prompt enhancement, single/multiple/concurrent generation, remix sequences, a rate-limit aware batch queue, and budget guardrails with a live “remaining budget” meter. It also includes download buttons for generated assets, optional FFmpeg stitching for multi-clip sequences, and one-click zip export of your session outputs.

Heads-up: The app is designed to be future-compatible. It uses a duration slider (1–60s) and a “Strict supported durations” toggle. When strict mode is on, your choice snaps to currently supported values (e.g., 4/8/12) to avoid API errors. You can turn strict mode off to try new durations as they become available.

https://sorastreamstudio.streamlit.app

Table of Contents

• Features
• Screenshots
• Requirements
• Quickstart
• Configuration
• Running on Streamlit Community Cloud
• Docker
• App Tour
  • Sidebar
  • Single Tab
  • Multiple Tab
  • Remix Tab
  • Batch Tab
  • Assets Tab
• Cost, Budget & Limits
• Reference Images
• Prompt Enhancement
• FFmpeg Stitching
• Troubleshooting
• Advanced Configuration
• Security Notes
• Roadmap
• Contributing
• License
• Buy Me a Coffee (Crypto)

---

Features

• Web UI (Streamlit) — zero-boilerplate local GUI for Sora video generation.
• Reference images:
  • Upload one or more images.
  • Or generate a reference image from a text prompt (uses an image model).
  • Automatic montage combining — upload multiple images and combine them into a single reference with configurable layouts (Auto, 1×N, N×1, 2×2, 3×2, 3×3, 3×N, Custom).
• Prompt enhancement (optional) with before/after display:
  • Styles: director, pixar, clean.
• Single generation flow.
• Multiple generation flow with concurrency:
  • Choose N jobs and workers.
  • Jobs run in parallel (threaded) and are still gated by a global rate limiter.
  • Row-wise, top-aligned prompt inputs for better UX.
• Remix flow:
  • Create a base clip, then apply N remix prompts sequentially.
  • Optional FFmpeg stitch into one MP4.
• Batch queue (local) with rate-limit aware scheduler:
  • Token-bucket limiter tied to your selected RPM tier.
  • Auto-pauses when tokens deplete and resumes after the 60-second refill window.
  • Optional enhancement for all prompts in the queue.
  • Shared reference images for the entire queue.
• Budget guardrails:
  • Set a session budget; see live spend meter and remaining budget.
  • Blocks new jobs that would exceed your cap.
• Cost estimator (upfront):
  • Based on model rates (defaults set in the code; update as needed).
• Assets tab:
  • Preview generated videos and references.
  • One-click download.
  • Zip session and clear outputs tools.

---

Requirements

• Python 3.10+ (3.11 recommended)
• OpenAI API key with Sora access
• FFmpeg (optional) for stitching clips:
  • macOS: brew install ffmpeg
  • Linux (Debian/Ubuntu): sudo apt-get install ffmpeg
  • Windows: Install from https://ffmpeg.org and add to PATH

---

Quickstart

1. Clone and install:

git clone https://github.com/Dzechy/sora-streamlit-studio.git
cd sora-streamlit-studio

# Minimal deps
pip install --upgrade streamlit openai pillow

1. Set your API key:

export OPENAI_API_KEY=sk-...    # macOS/Linux
# PowerShell (Windows):
# setx OPENAI_API_KEY "sk-..."

1. Run the app:

streamlit run Sora_Streamlit_Studio.py

Tip: You can also paste your API key directly into the app’s sidebar.

---

Configuration

Most settings live in the sidebar at runtime. For static defaults, edit the constants at the top of Sora_Streamlit_Studio.py:

PRICE_PER_SECOND = {"sora-2": 0.10, "sora-2-pro": 0.30}
CURRENT_SUPPORTED_DURATIONS = [4, 8, 12]   # informational hint for "Strict durations"
RPM_TIERS = {"Tier 1": 5, "Tier 2": 10, "Tier 3": 25, "Tier 4": 40, "Tier 5": 75}

If pricing, supported durations, or rate-limit tiers change, update them here.

---

Running on Streamlit Community Cloud

1. Push the repo to GitHub.
2. Create a new Streamlit app and target Sora_Streamlit_Studio.py.
3. Set Secrets:

• In your app dashboard → Settings → Secrets:

OPENAI_API_KEY = "sk-..."

1. Deploy. (You can also paste the key into the sidebar at runtime if you prefer.)

---

Docker (optional)

Dockerfile

FROM python:3.11-slim

# System deps for Pillow (optional) + ffmpeg (optional)
RUN apt-get update && apt-get install -y --no-install-recommends \
    ffmpeg && \
    rm -rf /var/lib/apt/lists/*

WORKDIR /app
COPY Sora_Streamlit_Studio.py /app/

RUN pip install --no-cache-dir streamlit openai pillow

ENV OPENAI_API_KEY=""
EXPOSE 8501

CMD ["streamlit", "run", "Sora_Streamlit_Studio.py", "--server.address=0.0.0.0", "--server.port=8501"]

Build & run:

docker build -t sora-studio .
docker run -it --rm -p 8501:8501 -e OPENAI_API_KEY=sk-... sora-studio

---

App Tour

Sidebar

API & Limits

• OpenAI API Key (masked) — or use OPENAI_API_KEY env var.
• Model: sora-2 or sora-2-pro.
• Duration (seconds): slider (1–60s).
• Strict supported durations: When enabled, your chosen value snaps to the nearest currently supported duration (listed beneath, e.g., 4/8/12) to avoid 400 errors. Turn it off to try new durations when the API updates.
• Frame Size: Presets or Custom (e.g., 720×1280).
• Rate limit tier (RPM): informs the token-bucket limiter that gates all API calls.

Prompt Enhancement

• Toggle on/off and pick a style: director, pixar, clean.

Reference Montage

• Montage layout: Choose from Auto, 1×N (vertical stack), N×1 (horizontal strip), 2×2 grid, 3×2 grid, 3×3 grid, 3×N (3 cols, rows as needed), or Custom.
• Padding: Set spacing between images in pixels.
• Custom columns/rows: For Custom layout, specify exact grid dimensions.

Budget

• Optional budget cap for the session, with live meter (spent / cap / remaining).
• Reset spend counter to zero the session’s totals.

Outputs

• Choose your output folder (session subfolder created automatically).
• New session folder, Zip session, Clear session.

---

Single tab

• Enter a video prompt.
• Optionally generate a reference image from text or upload multiple reference images to automatically combine them into a single montage.
• Optional prompt enhancement, with the enhanced version shown after generation.
• Estimated cost and API calls are shown before you run.
• On success, you’ll see a preview, download button, and (if used) the enhanced prompt.

---

Multiple tab

• Specify N generations and Concurrent workers.
• Provide N prompts using row-wise, top-aligned prompt inputs for easier editing.
• Optional shared reference image(s) (prompt-generated or uploaded, with automatic montage support).
• Optional enhancement per prompt; enhanced text appears with each result.
• Shows total estimated cost and total API calls up front.
• Global rate limiter still gates the calls to reduce 429s.

---

Remix tab

• Enter a Base Shot Prompt and a number of remix steps.
• Optional base reference images (prompt-generated or uploaded, with automatic montage support).
• Optional prompt enhancement (applies to base + each remix).
• Returns each clip with a download button.
• If FFmpeg is installed, you can stitch all the clips into one MP4.

---

Batch tab (rate-limit aware)

• Add any number of prompts to a local queue.
• When you Start queue run:
  • A token-bucket rate limiter paces API calls based on your RPM tier.
  • The scheduler auto-pauses when tokens run out and resumes when the minute window resets — no manual retry needed.
• You can enable prompt enhancement for the entire queue.
• Supply shared references for all batch jobs (generated or uploaded, with automatic montage support).
• The queue stops early if your budget cap would be exceeded by the next job.

---

Assets tab

• Shows all videos created in the current session folder and any reference images (uploaded or generated).
• One-click downloads.
• Works hand-in-hand with Zip session in the sidebar.

---

Cost, Budget & Limits

• Cost estimator uses per-second rates:
  • sora-2: $0.10/s
  • sora-2-pro: $0.30/s
• The app tracks spend after a clip successfully downloads (deduped by video id).
• Budget cap:
  • If enabled, the app warns when a planned run may exceed the remaining budget and blocks if it would.
• Rate limits:
  • The sidebar RPM setting configures a token-bucket limiter used by all API calls (enhancement, image generation, video create, remix).
  • This is advisory — pick the tier that fits your account to minimize 429s.

Update the constants at the top of Sora_Streamlit_Studio.py if pricing or rate-limit tiers change for your account.

---

Reference Images

You can either:

• Upload one or more images (PNG/JPG), or
• Generate a reference image from a short text prompt (uses an image model under the hood). The app saves it to your session’s refs folder and shows a preview.

Multiple images are automatically combined into a single reference montage using configurable layouts:

• Auto: Intelligently chooses grid layout based on image count
• 1×N / N×1: Vertical stack or horizontal strip
• 2×2 / 3×2 / 3×3: Fixed grid layouts
• 3×N: 3 columns with rows as needed
• Custom: Specify exact columns and rows

When generating a video, the app passes the combined reference image to the Sora API.

---

Prompt Enhancement

An optional one-click step that rewrites your prompt for clarity and specificity. Styles:

• director — cinematic language, camera, lensing, lighting, movement, environment
• pixar — warm, character-focused, wardrobe + blocking details
• clean — concise without changing intent

If enhancement fails (e.g., quota or rate-limit), the app falls back to your original prompt and continues.

---

FFmpeg Stitching

The Remix tab can stitch multiple clips into one MP4 if FFmpeg is found on your PATH.
Install FFmpeg (see Requirements) and you’ll see a “Stitch all clips” button once multiple clips exist.

---

Troubleshooting

• insufficient_quota / 429
  • If it’s quota, add billing or use a key with access.
  • If it’s burst rate-limit, reduce Concurrent workers or pick a lower RPM tier in the sidebar.
• Invalid seconds (400)
  • Enable Strict supported durations or select a supported duration (e.g., 4/8/12).
• Download failed
  • Check that your key has Sora permissions and that the job finished.
  • Intermittent issues are retried gently; try again if needed.
• FFmpeg not found
  • Install FFmpeg and restart the app; the stitch controls will appear.
• OpenAI SDK not installed
  • pip install openai — also ensure OPENAI_API_KEY is set or paste it into the sidebar.
• Pillow not installed
  • pip install pillow — required for reference image montage combining and resizing.

---

Advanced Configuration

Open Sora_Streamlit_Studio.py and adjust:

# Pricing (used by the cost estimator)
PRICE_PER_SECOND = {"sora-2": 0.10, "sora-2-pro": 0.30}

# Informational hint used by the "Strict supported durations" toggle
CURRENT_SUPPORTED_DURATIONS = [4, 8, 12]

# Rate-limit tiers for the token-bucket limiter
RPM_TIERS = {"Tier 1": 5, "Tier 2": 10, "Tier 3": 25, "Tier 4": 40, "Tier 5": 75}

You can also tweak:

• Default output root folder
• Default model list
• Retry counts / backoff behavior
• Thread pool workers for multi-generation
• Montage background color and padding defaults

---

Security Notes

• Never commit your API keys to source control.
• For Streamlit Cloud, store secrets in Settings → Secrets.
• Locally, prefer environment variables over hardcoding.

---

Roadmap

• Status dashboard (tokens remaining, next refill ETA)
• Per-job model/duration in Multi & Batch tabs
• Per-clip cost breakdown & CSV export
• Persistent job history across sessions
• Advanced montage options (custom backgrounds, borders, captions)

---

Contributing

PRs are welcome! Please:

1. Fork the repo.
2. Create a feature branch: feat/<your-feature>
3. Add tests or a test plan if applicable.
4. Submit a PR with a clear description and screenshots/GIFs.

---

License

MIT.

---

Buy Me a Coffee (Crypto)

If this saved you time (or a shoot 👀), you can fuel more dev with a small tip:

• Bitcoin (BTC): bc1qpav33sclt900r2q4q9y6vuewftx22jx0xun0pe
• Ethereum (ETH): 0x27d1FC34b431Aa9c80d5F9EE7464c03Eb1D15be0
• USDT (ETH Network): 0x27d1FC34b431Aa9c80d5F9EE7464c03Eb1D15be0

Thanks a ton! 🙏

---

Happy creating ✨