Generate Short Videos from Telegram Images: n8n + GPT-4.1 + Seedance/Veo3

Input: Telegram image with caption → Process: AI Agent generates video prompt and calls Seedance API → Output: Auto-delivered generated video back to Telegram

Who

Content creators or automation engineers who want to build a Telegram image-to-video pipeline in n8n and already have Google and Wavespeed credentials

You get

Users send an image and description to a Telegram Bot and automatically receive a short video generated by Seedance/Veo3

You need

n8n · telegram · google-drive · google-sheets

Steps

1. Configure credentials and import template

   Import template 6976 into n8n. Go to Credentials and add/configure: Telegram Bot, Google Drive, Google Sheets, OpenAI, and Wavespeed/Seedance (HTTP Header Auth). In the workflow's Google Drive and Google Sheets nodes, replace the REDACTED folderId and documentId with your actual IDs.

2. Set up Telegram Trigger and conditional branch

   Ensure the Telegram Trigger node has webhook enabled and listens for message updates. The If node checks whether $json.message.photo exists: if absent (no image), take the false branch; if present, proceed to the AI Agent branch.

3. Configure AI Agent for video prompt generation

   The AI Agent node uses the OpenAI Chat Model (default gpt-4.1-mini) and Simple Memory. The agent's input text is taken from the Telegram Trigger message text. The systemMessage instructs the model to generate a Seedance/Veo3-compatible video prompt and output JSON with two fields, prompt and image_url, where image_url is fetched from the last row of the attached Google Sheet tool.

   Execution prompt · Source text。AI Agent node Text input field

   Copy prompt

   ={{ $('Telegram Trigger').item.json.message.text }}

   Execution prompt · Source text。AI Agent node Options > System Message field

   Copy prompt

   =You are an assistant for video prompt and image-to-video generation.
   
   Your task is to:
   
   1. Create an effecgive Video Prompt to prompt a video generation model based on user's input
   
   2.Output only the following two values in JSON format:
   prompt: The text prompt for the video.
   image_url: The URL of the image to generate the video from. Fetch this value from the last row of the attached Google Sheet tool.
   Return only these two fields in your response, formatted as a JSON object.

4. Upload image to Google Drive and log to Sheets

   The Telegram node downloads the user image via file_id and passes it to the Google Drive node for upload to the specified folder, with filename 'Image Upload - {{ $now }}'. The Google Sheets node appends a new row recording the current date (Date) and Google Drive share link (Image Uploaded on Google Drive).

5. Call Wavespeed/Seedance API to generate video

   The Wavespeed Post node sends a POST request to https://api.wavespeed.ai/api/v3/bytedance/seedance-v1-pro-i2v-480p with body parameters: camera_fixed=false, duration=5, image={{ $json.message.content.image_url }}, prompt={{ $json.message.content.prompt }}, seed=-1. Uses HTTP Header Auth for authentication.

6. Poll video generation status and send back to Telegram

   After an initial 15-second wait, the Wavespeed Get node calls GET /predictions/{id}/result to check status. The If1 node checks if data.status equals 'completed': if complete, the Telegram2 node sends the video file back to the user via sendVideo; if not complete, it enters 'Wait another 15 Seconds' and polls again.

You are assisting with reproducing n8n template 6976: Telegram image-to-video workflow. Please first confirm the following are ready:
1. Access to an n8n instance;
2. Telegram Bot Token, Google OAuth credentials, OpenAI API Key, Wavespeed/Seedance API Key;
3. Google Drive target folder ID and Google Sheets document ID.
Once confirmed, proceed to import the template, replace placeholders, configure credentials, then activate the workflow and test by sending an image via Telegram.