Chapter 12: Image-to-Video Workflow with Google Drive and Telegram Integration¶

Video: Watch this chapter on YouTube (2:25:12)

Overview¶

This chapter builds a sophisticated image-to-video workflow that combines multiple triggers and integrations. When an image is uploaded to Google Drive, a Telegram notification is sent asking for video instructions. The user's response triggers video generation from the uploaded image.

Detailed Summary¶

Workflow Architecture (Dual Trigger)¶

This workflow has two distinct trigger paths in one worksheet:

Path 1: Image Upload Flow

Google Drive Trigger → Telegram (send photo) → Google Sheets (log image)

Path 2: Video Generation Flow

Telegram Trigger → Video Prompt Agent → Seedance POST → Wait → GET → If Loop → Telegram (send video)

Multi-Input Requirements¶

Image-to-video generation needs two inputs: 1. Base Image: The starting point for the video 2. Video Prompt: Description of desired video motion/action

Telegram serves as the communication hub for both inputs.

Part 1: Image Upload Detection¶

Step 1: Google Drive Trigger¶

Add Google Drive trigger node
Select "Watch for changes in a folder"
Choose your target folder (e.g., "Image to Video CodeCloud")
Watch for: "File created"
Test by uploading an image to the folder

Important: Make sure the folder/file is publicly accessible (or has proper sharing settings).

Key output: webContentLink - public URL to the uploaded image

Step 2: Telegram Bot Setup¶

Creating the Bot¶

Open Telegram
Search for BotFather (verify blue checkmark)
Start conversation: /start
Create new bot: /newbot
Enter bot name (e.g., "n8n Test Demo 1")
Enter bot username (must end in "bot")
Copy the API token provided

Connecting to n8n¶

Add Telegram node (not trigger)
Operation: Send Photo
Create new credential:
Paste API token
Click Save
Chat ID: Will be configured from trigger

Step 3: Getting Chat ID¶

Add Telegram Trigger node (separate trigger)
Select "On message"
Execute step (listening mode)
Send a message to your bot in Telegram
Check n8n for captured message
Copy the chat.id value

Use this chat ID for all Telegram nodes in the workflow.

Step 4: Configure Photo Notification¶

In Telegram send photo node:
Chat ID: Paste the chat ID (can be hardcoded)
Photo: Drag webContentLink from Google Drive
Caption: "You've uploaded this photo to the Google folder. Kindly provide the video idea you want to generate from this image."

Step 5: Log Image to Google Sheets¶

Add Google Sheets node
Operation: Append Row
Select your logging spreadsheet
Map fields:
Image URL: Drag webContentLink
Date: Drag $now variable

This creates a log of uploaded images for the video generation path to reference.

Part 2: Video Generation Flow¶

Step 1: Configure Telegram Response Handler¶

Use Telegram node with "Send message and wait for response":

Chat ID: Same as before
Message: "Could you provide the video idea you want to generate from this image?"
Response type: Free text

This pauses the workflow until the user responds.

Step 2: Video Prompt Agent¶

Add OpenAI node (Message Model)
User prompt: Drag text from Telegram response
System prompt: Video prompt engineer instructions
Add Google Sheets tool to fetch latest image URL

System Prompt Configuration¶

Your tasks:
1. Create an effective video prompt for a video generation model based on user input
2. Output two JSON objects:
   - "prompt": The generated video prompt (string)
   - "image_url": Fetch from last row of attached Google Sheets log tool

Enable "Output content as JSON" in node options.

Step 3: Seedance POST Request¶

Add HTTP Request node
Go to WaveSpeed → Models → Filter "image-to-video"
Select "Seedance V1"
Copy and import cURL

Body Configuration¶

{
  "prompt": "{{ $json.prompt }}",
  "image_url": "{{ $json.image_url }}",
  "duration": 5,
  "aspect_ratio": "16:9"
}

Step 4: Wait and Poll Pattern¶

Same as previous chapters: 1. Wait 15 seconds 2. GET request with request ID 3. If node checking status == completed 4. False branch: Wait 15s → Loop back to GET

Step 5: Telegram Video Delivery¶

Instead of Gmail, send video via Telegram:

Add Telegram node
Operation: Send Video
Chat ID: From trigger or hardcoded
Video: Drag output URL from If node

Complete Dual-Trigger Architecture¶

[Google Drive Trigger]              [Telegram Trigger]
        ↓                                   ↓
[Telegram: Send Photo]            [Telegram: Wait for response]
        ↓                                   ↓
[Google Sheets: Log]              [Video Prompt Agent + Sheets Tool]
        ↓                                   ↓
   [Execute                        [Seedance POST]
   Subworkflow] ←--optional--→            ↓
                                  [Wait 15s]
                                          ↓
                                  [GET Request]
                                          ↓
                                  [If: completed?]
                                    ├── True → [Telegram: Send Video]
                                    └── False → [Wait] → Loop

Best Practice: Split into Subworkflows¶

The chapter notes that having two triggers in one workflow can complicate troubleshooting. Consider:

Workflow 1: Image upload → Notification → Log
Workflow 2: Message trigger → Video generation → Delivery

Use Execute Subworkflow node to connect them.

End-to-End Testing¶

Upload image to Google Drive folder
Receive Telegram notification with image
Reply with video idea
Wait for video generation
Receive video in Telegram

Alternative Configurations¶

Replace Seedance with VO3, Kling, or other models
Use Gmail instead of Telegram for delivery
Add Slack notifications
Store videos in Google Drive instead of sending links

Key Takeaways¶

Multi-trigger workflows are possible: Two separate trigger paths can coexist in one workflow.
Telegram enables interactive workflows: Send-and-wait creates conversational automation.
Google Sheets as a bridge: Log data in sheets, then retrieve in another path using tools.
BotFather creates Telegram bots: Simple setup for custom automation bots.
Chat ID is essential: Every Telegram operation needs the correct chat ID.
Image-to-video needs both inputs: Base image + motion description.
JSON output mode helps: When multiple values needed, output as structured JSON.
Consider splitting workflows: Dual triggers can be separated for easier maintenance.
Publicly accessible images required: Video APIs need accessible image URLs.
Same polling pattern applies: POST → Wait → GET → Check → Loop or deliver.

Conclusion¶

The image-to-video workflow showcases n8n's ability to orchestrate complex, multi-step processes across multiple services. By combining Google Drive for image intake, Telegram for user interaction, Google Sheets for data persistence, and WaveSpeed for video generation, the workflow demonstrates enterprise-level automation patterns. The dual-trigger architecture, while functional, also illustrates the importance of workflow organization—a theme expanded in the subworkflow chapter. This real-world use case applies to marketing automation, content creation pipelines, and any scenario requiring human input mid-workflow.