🦖🎬 Seedance 2.0 ComfyUI Node — Sjinn.ai API

Generate stunning videos with ByteDance Seedance 2.0 directly in ComfyUI via Sjinn.ai

"The most convenient way to access ByteDance Seedance 2.0 directly inside ComfyUI — connect your reference images, write a prompt, click Run."

Author: Voogie | Project: Cameraptor | cameraptor.com/voogie

---

Example workflow: 2 reference images + 1 reference video → Seedance 2.0 → output

---

🤔 What is this?

Sjinn.ai is an AI platform giving Pro+ members API access to ByteDance's Seedance 2.0 — one of the most capable video generation models available. This ComfyUI node wraps the Sjinn.ai API so you can run Seedance 2.0 generations directly in your workflow: connect reference images and videos, write a prompt, click Run. Cost: 100 credits per second of output video — a 5-second video costs 500 credits.

---

✨ Features

Feature	Details
🖼️ Reference images	Up to 6 — connect any Load Image node to image_1 through image_6
🎥 Reference videos	Up to 3 — connect the vanilla ComfyUI Load Video node to video_1 through video_3
📐 Aspect ratios	16:9, 9:16, 1:1, 4:3, 3:4
⏱️ Duration	4 to 15 seconds (100 credits per second)
🎞️ First frame preview	IMAGE output — pipe into upscaler, preview node, or any image workflow
📊 Real-time console log	Progress printed every 15 seconds with elapsed time and attempt count

---

📦 Prerequisites

• ComfyUI (v0.14+ recommended — uses native VIDEO type)
• Sjinn.ai Pro+ account — https://sjinn.ai (required for API access)
• ComfyUI Manager (for easy installation)
• ComfyUI-VideoHelperSuite — required for the included example workflow (Seedance_2_By_Voogie.json)
• opencv-python — for first_frame extraction (auto-installed via requirements.txt)

💡 The vanilla Load Video node ships with ComfyUI v0.14+ — no extra install needed for video inputs.

---

🚀 Installation

Method 1: Via ComfyUI Manager ⭐ (Recommended)

1. Launch ComfyUI and open the Manager
2. Click Install via Git URL
3. Paste the repository URL:

   https://github.com/Cameraptor/seedance_2_Comfy_UI_Node-sjinn_Api-
   

4. Click Install (dependencies will be installed automatically)
5. Restart ComfyUI completely

Method 2: Manual Installation

1. Navigate to your ComfyUI/custom_nodes/ directory and clone:

   git clone https://github.com/Cameraptor/seedance_2_Comfy_UI_Node-sjinn_Api-
   cd seedance_2_Comfy_UI_Node-sjinn_Api-

2. Install dependencies using ComfyUI's Python:

   Windows (portable):

   ..\..\python_embeded\python.exe -m pip install -r requirements.txt

   Linux / Mac:

   ../../python/bin/python -m pip install -r requirements.txt

3. Restart ComfyUI completely

---

🔑 Getting Your Credentials

This is the most important section — the node requires two separate credentials from Sjinn.ai. Both are required.

API Key

1. Go to https://sjinn.ai/ and log in
2. Click your profile avatar (top right)
3. Navigate to the API Keys section
4. Click Create new key, give it a name
5. Copy the key immediately (shown only once)
6. Paste into the api_key field of the Seedance 2.0 node

⚠️ The API key is used only for polling task status. You also need a session token (see below) for file uploads and task creation — they use different authentication.

Session Token

The session token is a browser cookie that Sjinn.ai uses to authenticate file uploads and video generation requests. You need to grab it from your browser's DevTools — here's exactly how:

Step 1 — Open the page and log in

Go to https://sjinn.ai/tools/seedance20-video in your browser and make sure you are logged in with a Pro+ account.

Step 2 — Open Developer Tools

Press F12 on your keyboard (or right-click anywhere on the page → Inspect). A panel will open at the bottom or side of your browser window.

Step 3 — Navigate to the cookies

• Chrome / Edge: Click the Application tab at the top of the DevTools panel
• Firefox: Click the Storage tab

Then in the left sidebar, expand the tree:

▼ Storage
  ▼ Cookies
      https://sjinn.ai   ← click this

💡 If you don't see the Application / Storage tab, click the » arrow at the end of the tab bar — it may be hidden.

Step 4 — Find the right cookie

A table of cookies will appear on the right. Look for the one named:

__Secure-next-auth.session-token

⚠️ There are many cookies in the list — you need exactly this one. Use Ctrl+F to search if the list is long.

Step 5 — Copy the value

Click on that cookie row. At the bottom of the DevTools panel you'll see the full value — it's a very long string that starts with eyJ...

To copy it correctly:

1. Click anywhere in the value field at the bottom
2. Press Ctrl+A to select all (the value is too long to see fully in the table row)
3. Press Ctrl+C to copy

⚠️ Don't copy from the table row — the value is truncated there. Always copy from the bottom preview panel where the full string is shown.

Step 6 — Paste into the node

Paste the copied value into the session_token field of the Seedance 2.0 node in ComfyUI. The value should start with eyJ and be several hundred characters long — if it's short, something went wrong.

---

⚠️ Token expiry: The session token is valid as long as you stay logged into Sjinn.ai (typically days to weeks). If you get 401 errors, your session expired — log in again and repeat the steps above to get a fresh token.

💡 Why not just use the API key for everything? Sjinn.ai's video creation and upload endpoints only accept the session cookie — Bearer API keys return 401 there. The API key is used only for polling task status. Both credentials are required.

---

🕹️ How to Use

Find the 🦖🎬 Seedance 2.0 node in the 🦖Cameraptor_Nodes / Sjinn.ai menu, or search "Seedance" in ComfyUI node search.

1. Connect reference images: Load Image → image_1 (up to 6 total)
2. Connect reference videos: Load Video → video_1 (up to 3 total)
3. Write your prompt — use @Image1, @Image2, @Video1 etc. to reference your inputs
   • Example: @Image1 is FIRST FRAME. @Image2 is the vehicle reference. @Video1 is camera motion reference.
4. Set ratio and duration
5. Enter your session_token and api_key
6. Click Queue Prompt — the node uploads assets, submits the task, and polls every 15 seconds
7. When complete: video_url output contains the direct MP4 link

💡 Connect video_url to any save or download node. Connect first_frame to a Preview Image node to see the result instantly without downloading.

---

📋 Node Parameters

Inputs — Required:

Parameter	Type	Description
session_token	STRING	NextAuth browser cookie from sjinn.ai (see Getting Credentials)
api_key	STRING	Sjinn.ai API key — used for status polling only
prompt	STRING	Text prompt. Use @Image1..@Image6, @Video1..@Video3 to reference inputs

Inputs — Optional:

Parameter	Type	Description
image_1 .. image_6	IMAGE	Reference images — connect Load Image nodes
video_1 .. video_3	VIDEO	Reference videos — connect vanilla Load Video node
ratio	COMBO	Output aspect ratio: 16:9 / 9:16 / 1:1 / 4:3 / 3:4 (default: 16:9)
duration	COMBO	Output duration in seconds: 4–15 (default: 5)

Outputs:

Output	Type	Description
video_url	STRING	Direct HTTPS URL to the generated MP4 file
info	STRING	Summary: task_id, elapsed time, ratio, duration, asset counts
first_frame	IMAGE	First frame of the video — for preview or chaining into image workflows

---

💰 Cost Reference

Duration	Credits Used
4s	400 credits
5s	500 credits
10s	1,000 credits
15s	1,500 credits

💡 Check your credit balance at https://sjinn.ai before running long generations.

✅ Content policy: If your prompt violates platform rules, the task returns status=2 (failed) and credits are automatically refunded.

---

💡 Prompt Tips & Best Practices

• Use @Image1, @Image2... to explicitly tell the model which image is which
• Put the most important reference first (image_1)
• @Video1 works well for camera motion and composition reference
• Avoid brand names in prompts — may trigger content filters (credits refunded if so)
• For character consistency: put character reference in @Image2, scene in @Image1
• Duration 4–5s generates faster and costs less — good for testing prompts before running full duration

---

📦 Example Workflow

The file Seedance_2_By_Voogie.json is an example workflow included in this repository.

To load: Drag the JSON file onto the ComfyUI canvas, or use Load from the menu.

The workflow demonstrates:

• 2× Load Image → Seedance 2.0 (image_1 + image_2)
• 1× Load Video → Seedance 2.0 (video_1)
• Seedance 2.0 → 🦖💾 Save API Video — downloads the MP4 and loads frames
• Seedance 2.0 first_frame → Preview Image (instant thumbnail in ComfyUI)
• Save API Video frames → VHS Video Combine — preview video in ComfyUI

⚠️ Requirements for the example workflow:

• ComfyUI-VideoHelperSuite — for the VHS Video Combine node. Install via ComfyUI Manager → search "VideoHelperSuite".

---

🛠️ Troubleshooting

Issue	Likely Cause	Fix
401 Unauthorized during upload or task creation	Session token expired or invalid	Log into sjinn.ai, get a fresh cookie value from DevTools
Read timed out (task creation)	sjinn.ai server slow to respond	Retry — this is not a code bug and usually resolves itself
Task failed: status=2	Content policy violation in prompt	Remove brand names, soften motion descriptions
video_1 socket not visible	Old ComfyUI version (pre VIDEO type)	Update ComfyUI to v0.14+
No videos in Load Video dropdown	No .mp4 files in ComfyUI/input/	Copy .mp4 files into ComfyUI/input/, then refresh
first_frame is black 64×64	opencv-python not installed	Run: pip install opencv-python
Task hangs for more than 15 minutes	Server-side issue or timeout	Check sjinn.ai status, retry the generation

---

🔐 Optional: Local Credentials File

To avoid re-entering credentials every session, create sjinn_config.py in the node folder:

SJINN_API_KEY = "your-api-key-here"
SJINN_SESSION_TOKEN = "your-session-token-here"

The node loads it automatically if present, falls back to empty fields if not.

⚠️ Security: Never commit this file to git. It is listed in .gitignore — keep it that way.

---

📝 Requirements

• ComfyUI installed and running (v0.14+ recommended)
• Sjinn.ai Pro+ account with credits
• API key from sjinn.ai profile
• Session token from browser DevTools
• Python packages installed (pip install -r requirements.txt)
• opencv-python installed (for first_frame output)

---

🤝 Support & Community

• 💬 Telegram Community: Join @voogieboogie — Get help, share workflows, discuss features!
• 🐛 Issues: GitHub Issues
• 🌐 Website: cameraptor.com/voogie

---

📄 License

MIT License — see LICENSE file for details.

---

Made with ❤️ by Voogie | Cameraptor

⭐ Star this repo if you find it useful!