klbr/n8nworkflows.xyz

Fork 0

mirror of https://github.com/khoaliber/n8nworkflows.xyz.git synced 2026-04-19 17:14:37 +00:00

Files

nusquama ee43b6f730 creation

2026-03-15 12:01:56 +08:00

46 KiB

Raw Blame History

Generate 360° product videos from photos with Veo 3 and Telegram

https://n8nworkflows.xyz/workflows/generate-360--product-videos-from-photos-with-veo-3-and-telegram-13928

Generate 360° product videos from photos with Veo 3 and Telegram

1. Workflow Overview

This workflow turns a product photo sent to a Telegram bot into a generated 360° product video using Google Vertex AI Veo 3, then sends the resulting video back to the same Telegram user.

Its main use case is product visualization: a user submits a single image, and the workflow produces a cinematic orbit-style showcase video. The process includes input validation, Google Cloud authentication via a service account stored in Google Sheets, image preparation, Veo 3 submission, polling for completion, and Telegram delivery.

1.1 Input Reception and Validation

The workflow starts from a Telegram bot message. It verifies that the incoming message contains a photo and that the image meets a minimum size requirement.

1.2 Google Cloud Authentication

The workflow reads Google service account fields from Google Sheets, locally signs a JWT in a Code node, and exchanges that JWT for a Google OAuth access token.

1.3 Image Download and Conversion

Once authenticated, the workflow downloads the Telegram-hosted image file and converts it into Base64 so it can be embedded in the Vertex AI request payload.

1.4 Veo 3 Job Submission

The workflow builds a prompt and request payload for Veo 3, submits a long-running generation job to Vertex AI, and captures the returned operation name.

1.5 Polling and Timeout Handling

The workflow waits 2 minutes between checks and repeatedly polls Vertex AI until the operation is done or a timeout threshold is reached.

1.6 Video Delivery

When the video bytes are available, the workflow converts them into a binary file and sends the video back to the Telegram user.

2. Block-by-Block Analysis

2.1 Input Reception and Validation

Overview:
This block listens for Telegram messages and verifies that the user sent a usable photo. It extracts core metadata such as chat ID, message ID, caption, and file ID for downstream processing.

Nodes Involved:

Telegram Trigger
1. Validate Input

Node: Telegram Trigger

Type and technical role: n8n-nodes-base.telegramTrigger — entry point that receives Telegram message updates.
Configuration choices:
- Configured to listen only to message updates.
- Uses Telegram API credentials tied to the bot.
Key expressions or variables used: None in node parameters.
Input and output connections:
- No input; this is a workflow entry point.
- Outputs to 2. Validate Input.
Version-specific requirements: Type version 1.2.
Edge cases or potential failure types:
- Telegram webhook misconfiguration
- Invalid Telegram credentials
- Bot not enabled or blocked by user
- Updates other than standard messages are ignored
Sub-workflow reference: None.

Node: 2. Validate Input

Type and technical role: n8n-nodes-base.code — custom JavaScript validation and normalization.
Configuration choices:
- Reads message from the Telegram Trigger payload.
- Fails if no message exists.
- Fails if no photo array exists.
- Uses the largest Telegram photo variant from message.photo[message.photo.length - 1].
- Rejects images where the smaller dimension is below 480 pixels.
- Extracts:
  - chatId
  - messageId
  - caption
  - file_id
  - error
  - errorMessage if validation fails
Key expressions or variables used:
- message.photo[message.photo.length - 1]
- Math.min(photo.width, photo.height) < 480
Input and output connections:
- Input from Telegram Trigger
- Output to 1. Get Service Account Details
Version-specific requirements: Type version 2.
Edge cases or potential failure types:
- Text-only messages
- Files sent as Telegram documents rather than photos
- Missing chat or message_id
- Very small images
Sub-workflow reference: None.

Important implementation note:
Although the sticky note says invalid input is rejected with a clear error, the current connection design does not branch on error immediately after validation. Instead, the workflow continues into the auth block. As built, validation errors may propagate until later and may not stop early in a clean way.

2.2 Google Cloud Authentication

Overview:
This block retrieves service account fields from Google Sheets, signs a JWT using the private key, exchanges it for an OAuth token, and checks whether token retrieval succeeded.

Nodes Involved:

1. Get Service Account Details
1. Build JWT from Sheet
1. Get Access Token
Check Auth Token Valid
Send Validation Error
Send Processing Message

Node: 1. Get Service Account Details

Type and technical role: n8n-nodes-base.googleSheets — fetches service account credential fields from a Google Sheet.
Configuration choices:
- Reads from a spreadsheet placeholder identified as YOUR_GOOGLE_SHEET_ID_HERE.
- Sheet selected as gid=0.
- Uses Google Sheets OAuth2 credentials.
Key expressions or variables used: None visible in parameters.
Input and output connections:
- Input from 2. Validate Input
- Output to 2. Build JWT from Sheet
Version-specific requirements: Type version 4.7.
Edge cases or potential failure types:
- Spreadsheet ID not replaced
- OAuth permission issues
- Missing expected columns: client_email, private_key, project_id, scope
- Empty sheet or multiple unexpected rows
Sub-workflow reference: None.

Node: 2. Build JWT from Sheet

Type and technical role: n8n-nodes-base.code — creates a signed JWT assertion from sheet-provided service account fields.
Configuration choices:
- Imports Node.js crypto.
- Uses:
  - client_email
  - private_key
  - project_id
  - scope with default https://www.googleapis.com/auth/cloud-platform
- Replaces escaped newlines in the private key.
- Builds JWT header and claims.
- Signs with RSA-SHA256.
- Returns jwt, project_id, client_email, and error state.
Key expressions or variables used:
- item.private_key.replace(/\\n/g, '\n')
- now + 3600 for token lifetime
Input and output connections:
- Input from 1. Get Service Account Details
- Output to 3. Get Access Token
Version-specific requirements: Type version 2; requires Code node support for require('crypto').
Edge cases or potential failure types:
- Malformed PEM private key
- Missing service account fields
- Restricted Code-node environment
- Signature generation errors
Sub-workflow reference: None.

Node: 3. Get Access Token

Type and technical role: n8n-nodes-base.httpRequest — exchanges JWT for a Google OAuth2 access token.
Configuration choices:
- POST https://oauth2.googleapis.com/token
- Form URL encoded body:
  - grant_type=urn:ietf:params:oauth:grant-type:jwt-bearer
  - assertion={{ $json.jwt }}
Key expressions or variables used:
- ={{ $json.jwt }}
Input and output connections:
- Input from 2. Build JWT from Sheet
- Output to Check Auth Token Valid
Version-specific requirements: Type version 4.2.
Edge cases or potential failure types:
- Invalid JWT
- Expired or malformed claims
- Incorrect OAuth scope
- HTTP 400/401 responses
Sub-workflow reference: None.

Node: Check Auth Token Valid

Type and technical role: n8n-nodes-base.if — checks whether access_token is present and non-empty.
Configuration choices:
- Condition evaluates ={{ $json.access_token.isEmpty() }} and checks that result is false.
- True branch means token is considered valid.
- False branch sends an error.
Key expressions or variables used:
- ={{ $json.access_token.isEmpty() }}
Input and output connections:
- Input from 3. Get Access Token
- True output to Send Processing Message
- False output to Send Validation Error
Version-specific requirements: Type version 2.
Edge cases or potential failure types:
- If access_token is undefined, .isEmpty() may be brittle depending on runtime behavior
- Upstream error payloads may not include chatId and messageId, making error messaging unreliable
Sub-workflow reference: None.

Node: Send Validation Error

Type and technical role: n8n-nodes-base.telegram — sends an error message back to the user.
Configuration choices:
- Text: ={{ $json.errorMessage }}
- Chat ID: ={{ $json.chatId }}
- Replies to original message via messageId
Key expressions or variables used:
- ={{ $json.errorMessage }}
- ={{ $json.chatId }}
- ={{ $json.messageId }}
Input and output connections:
- Input from false branch of Check Auth Token Valid
- No downstream nodes
Version-specific requirements: Type version 1.1.
Edge cases or potential failure types:
- Missing chatId or messageId
- Telegram send failure
- Empty errorMessage
Sub-workflow reference: None.

Node: Send Processing Message

Type and technical role: n8n-nodes-base.telegram — acknowledges processing has started.
Configuration choices:
- Sends a user-facing status message:
  - “Creating your 360° product video...”
  - Estimated 3–4 minutes
  - Indicates Google Veo 3 processing
- Replies to original message
Key expressions or variables used:
- ={{ $json.chatId }}
- ={{ $json.messageId }}
Input and output connections:
- Input from true branch of Check Auth Token Valid
- Output to Download Image
Version-specific requirements: Type version 1.1.
Edge cases or potential failure types:
- Missing chatId or messageId
- Telegram send failure
Sub-workflow reference: None.

Important implementation note:
There is a data continuity issue in this block: 1. Get Service Account Details likely outputs only sheet row data and may overwrite fields created by 2. Validate Input, such as chatId, messageId, and caption, unless the node is configured to preserve or merge incoming fields. Several later nodes assume those values still exist. This should be verified in n8n runtime.

2.3 Image Download and Conversion

Overview:
This block downloads the selected Telegram image variant and converts the binary file into Base64 with MIME metadata for Vertex AI. If conversion fails, the user is notified.

Nodes Involved:

Download Image
Convert Image to Base64
Conversion OK?
Send Conversion Error

Node: Download Image

Type and technical role: n8n-nodes-base.telegram — downloads a Telegram file.
Configuration choices:
- Resource: file
- fileId is set to ={{ $json.result.reply_to_message.photo[3].file_id }}
Key expressions or variables used:
- ={{ $json.result.reply_to_message.photo[3].file_id }}
Input and output connections:
- Input from Send Processing Message
- Output to Convert Image to Base64
Version-specific requirements: Type version 1.1.
Edge cases or potential failure types:
- This expression depends on the output structure of Send Processing Message
- It assumes exactly at least 4 photo sizes exist
- It ignores the previously extracted file_id
- Telegram API response shape may not include result.reply_to_message.photo
Sub-workflow reference: None.

Important implementation note:
This is one of the riskiest nodes in the workflow. A safer expression would typically use the validated file ID from upstream data, such as {{$json.file_id}}, rather than reply_to_message.photo[3].file_id.

Node: Convert Image to Base64

Type and technical role: n8n-nodes-base.code — converts binary image data to Base64 and normalizes MIME handling.
Configuration choices:
- Iterates over all input items.
- Tries multiple binary data shapes:
  - nested .data
  - Buffer
  - string
  - fallback through binary keys
- Produces:
  - imageBase64
  - imageMimeType
  - imageSize
  - conversionStatus
- On failure produces:
  - error
  - conversionStatus: failed
  - conversionError
Key expressions or variables used:
- item.binary.data
- fallback inspection through Object.keys(item.binary)
Input and output connections:
- Input from Download Image
- Output to Conversion OK?
Version-specific requirements: Type version 2.
Edge cases or potential failure types:
- Telegram file download produced no binary data
- Unexpected binary property name
- Empty image buffer
- MIME metadata missing or inaccurate
Sub-workflow reference: None.

Node: Conversion OK?

Type and technical role: n8n-nodes-base.if — branches on conversion success.
Configuration choices:
- Compares conversionStatus to success
Key expressions or variables used:
- ={{ $json.conversionStatus }}
Input and output connections:
- Input from Convert Image to Base64
- True output to 5. Prepare Veo Request
- False output to Send Conversion Error
Version-specific requirements: Type version 1.
Edge cases or potential failure types:
- Missing conversionStatus
Sub-workflow reference: None.

Node: Send Conversion Error

Type and technical role: n8n-nodes-base.telegram — informs the user that image processing failed.
Configuration choices:
- Sends detailed error with conversionError
- Replies to original message
Key expressions or variables used:
- =❌ Failed to process your image: {{ $json.conversionError }}
- ={{ $json.chatId }}
- ={{ $json.messageId }}
Input and output connections:
- Input from false branch of Conversion OK?
- No downstream nodes
Version-specific requirements: Type version 1.1.
Edge cases or potential failure types:
- Missing chatId, messageId, or conversionError
Sub-workflow reference: None.

2.4 Veo 3 Job Submission

Overview:
This block assembles the generation prompt and Veo request body, submits the request to Vertex AI, and extracts the operation name used for asynchronous polling.

Nodes Involved:

1. Prepare Veo Request
1. Call Vertex AI Veo 3
Extract Operation Name

Node: 5. Prepare Veo Request

Type and technical role: n8n-nodes-base.code — builds the model prompt and API payload.
Configuration choices:
- Base prompt requests:
  - 360-degree product showcase
  - smooth orbit camera
  - clean white background
  - centered product
  - cinematic movement
- If Telegram caption exists, appends it as product context.
- Payload includes:
  - aspectRatio: 16:9
  - sampleCount: 1
  - durationSeconds: 8
  - personGeneration: allow_all
  - addWatermark: true
  - includeRaiReason: true
  - generateAudio: true
- Returns:
  - api_payload as a JSON string
  - chatId
  - messageId
  - caption
Key expressions or variables used:
- item.caption || ''
- JSON.stringify(apiPayload)
Input and output connections:
- Input from Conversion OK? true branch
- Output to 6. Call Vertex AI Veo 3
Version-specific requirements: Type version 2.
Edge cases or potential failure types:
- Missing imageBase64
- Very large payload size
- Unsupported MIME type
- Caption containing prompt-injection-like content
Sub-workflow reference: None.

Node: 6. Call Vertex AI Veo 3

Type and technical role: n8n-nodes-base.httpRequest — submits a long-running Veo generation request.
Configuration choices:
- POST to: https://us-central1-aiplatform.googleapis.com/v1/projects/{{project_id}}/locations/us-central1/publishers/google/models/veo-3.0-generate-preview:predictLongRunning
- Sends JSON body from api_payload
- Headers:
  - Authorization: Bearer {{ access_token }}
  - Content-Type: application/json
Key expressions or variables used:
- {{ $('3. Get Access Token').item.json.project_id }}
- {{ $('3. Get Access Token').item.json.access_token }}
- ={{ $json.api_payload }}
Input and output connections:
- Input from 5. Prepare Veo Request
- Output to Extract Operation Name
Version-specific requirements: Type version 4.2.
Edge cases or potential failure types:
- Veo preview access not enabled
- Invalid project or region
- Invalid access token
- Request schema mismatch
- Quota, safety, or model availability errors
Sub-workflow reference: None.

Node: Extract Operation Name

Type and technical role: n8n-nodes-base.code — extracts the long-running operation name from Vertex AI response.
Configuration choices:
- If response.name exists, returns:
  - operation_name
  - status: polling
  - poll_count: 0
  - chatId
  - messageId
  - caption
- Else returns failure information.
Key expressions or variables used:
- response.name
Input and output connections:
- Input from 6. Call Vertex AI Veo 3
- Output to Wait 2 Minutes
Version-specific requirements: Type version 2.
Edge cases or potential failure types:
- Vertex AI returns an error object instead of name
- Upstream HTTP node may not preserve request context fields
Sub-workflow reference: None.

Important implementation note:
Extract Operation Name expects chatId, messageId, and caption to exist on the HTTP response item. Depending on the HTTP Request node behavior, these values may not be preserved unless explicitly included or merged.

2.5 Polling and Timeout Handling

Overview:
This block waits two minutes between status checks, polls the long-running Vertex AI operation, and either proceeds to delivery when the video is ready or loops again until a timeout occurs.

Nodes Involved:

Wait 2 Minutes
Poll Video Status
Is Video Ready?
Continue Polling
Check Timeout
Send Timeout Error

Node: Wait 2 Minutes

Type and technical role: n8n-nodes-base.wait — delay between polling attempts.
Configuration choices:
- Wait duration: 2 minutes
Key expressions or variables used: None.
Input and output connections:
- Input from Extract Operation Name
- Also loops back from Check Timeout false branch
- Output to Poll Video Status
Version-specific requirements: Type version 1.1.
Edge cases or potential failure types:
- Wait execution persistence depends on n8n setup
- If wait/resume webhooks are not supported in the hosting setup, resumption can fail
Sub-workflow reference: None.

Node: Poll Video Status

Type and technical role: n8n-nodes-base.httpRequest — checks operation status with Vertex AI.
Configuration choices:
- POST to: https://us-central1-aiplatform.googleapis.com/v1/projects/{{project_id}}/locations/us-central1/publishers/google/models/veo-3.0-generate-preview:fetchPredictOperation
- JSON body:
  - operationName: {{ $json.operation_name }}
Key expressions or variables used:
- {{ $('3. Get Access Token').item.json.project_id }}
- {{ $json.operation_name }}
Input and output connections:
- Input from Wait 2 Minutes
- Output to Is Video Ready?
Version-specific requirements: Type version 4.2.
Edge cases or potential failure types:
- Missing Authorization header: this node currently does not explicitly set one
- Expired token if the workflow runs too long
- Invalid or missing operation_name
Sub-workflow reference: None.

Important implementation note:
Unlike the submission node, this polling node does not define an Authorization header. If the Vertex AI polling endpoint requires OAuth, this node will fail unless credentials are injected elsewhere or the API accepts unauthenticated polling, which is unlikely.

Node: Is Video Ready?

Type and technical role: n8n-nodes-base.if — checks whether the operation is complete.
Configuration choices:
- Condition: done === true
Key expressions or variables used:
- ={{ $json.done }}
Input and output connections:
- Input from Poll Video Status
- True output to Convert Video to File
- False output to Continue Polling
Version-specific requirements: Type version 2.
Edge cases or potential failure types:
- Missing done property
- Response schema changes from Vertex AI
Sub-workflow reference: None.

Node: Continue Polling

Type and technical role: n8n-nodes-base.code — increments poll count and decides whether timeout is reached.
Configuration choices:
- Increments poll_count
- If pollCount > 40, returns timeout state
- Otherwise returns:
  - operation_name
  - poll_count
  - status: polling
  - chatId
  - messageId
  - caption
Key expressions or variables used:
- (item.poll_count || 0) + 1
- item.name || item.operation_name
Input and output connections:
- Input from false branch of Is Video Ready?
- Output to Check Timeout
Version-specific requirements: Type version 2.
Edge cases or potential failure types:
- Poll count may be lost if HTTP responses do not preserve original JSON fields
- operation_name may be missing if relying on item.name
Sub-workflow reference: None.

Node: Check Timeout

Type and technical role: n8n-nodes-base.if — checks whether the loop should stop with timeout.
Configuration choices:
- Condition: status == "timeout"
Key expressions or variables used:
- ={{ $json.status }}
Input and output connections:
- Input from Continue Polling
- True output to Send Timeout Error
- False output to Wait 2 Minutes
Version-specific requirements: Type version 2.
Edge cases or potential failure types:
- Missing status
Sub-workflow reference: None.

Node: Send Timeout Error

Type and technical role: n8n-nodes-base.telegram — sends timeout notification to the user.
Configuration choices:
- Message includes the number of attempts from poll_count
- Replies to original message
Key expressions or variables used:
- {{ $json.poll_count }}
- ={{ $json.chatId }}
- ={{ $json.messageId }}
Input and output connections:
- Input from true branch of Check Timeout
- No downstream nodes
Version-specific requirements: Type version 1.1.
Edge cases or potential failure types:
- Missing chatId, messageId, or poll_count
Sub-workflow reference: None.

2.6 Video Delivery

Overview:
This block converts the Base64 video returned by Vertex AI into a binary file and sends it to the Telegram chat.

Nodes Involved:

Convert Video to File
Send Video to User

Node: Convert Video to File

Type and technical role: n8n-nodes-base.convertToFile — converts Base64 video bytes into binary output.
Configuration choices:
- Operation: toBinary
- Source property: response.videos[0].bytesBase64Encoded
Key expressions or variables used:
- response.videos[0].bytesBase64Encoded
Input and output connections:
- Input from true branch of Is Video Ready?
- Output to Send Video to User
Version-specific requirements: Type version 1.1.
Edge cases or potential failure types:
- Missing response.videos[0]
- Unexpected Vertex AI output schema
- Empty or invalid Base64 data
Sub-workflow reference: None.

Node: Send Video to User

Type and technical role: n8n-nodes-base.telegram — sends the generated video as Telegram media.
Configuration choices:
- Operation: sendVideo
- Uses binary data
- Chat ID is taken from the original trigger: ={{ $('Telegram Trigger').item.json.message.chat.id }}
Key expressions or variables used:
- ={{ $('Telegram Trigger').item.json.message.chat.id }}
Input and output connections:
- Input from Convert Video to File
- No downstream nodes
Version-specific requirements: Type version 1.1.
Edge cases or potential failure types:
- Cross-node item reference may fail if execution context changes after Wait node resume
- Large file may exceed Telegram limits
- Missing binary property name if Convert to File uses a different default output key
Sub-workflow reference: None.

Important implementation note:
This node references the original Telegram Trigger item directly instead of using chatId carried through the workflow. After a Wait node resume, direct item references can be fragile. Using {{$json.chatId}} would usually be more reliable.

3. Summary Table

Node Name	Node Type	Functional Role	Input Node(s)	Output Node(s)	Sticky Note
Telegram Trigger	Telegram Trigger	Receives Telegram message updates from the bot		2. Validate Input	## Receive & Validate Listens for Telegram messages and checks that a photo was sent and is at least 480px. Rejects documents or text-only messages and replies with a clear error.
2. Validate Input	Code	Validates message content and extracts photo/chat metadata	Telegram Trigger	1. Get Service Account Details	## Receive & Validate Listens for Telegram messages and checks that a photo was sent and is at least 480px. Rejects documents or text-only messages and replies with a clear error.
1. Get Service Account Details	Google Sheets	Reads service account fields from a Google Sheet	2. Validate Input	2. Build JWT from Sheet	## Google Cloud Auth Reads Service Account credentials from Google Sheets, signs a JWT locally, and exchanges it for a short-lived OAuth access token. Runs fresh on every request.
2. Build JWT from Sheet	Code	Builds and signs a Google service account JWT locally	1. Get Service Account Details	3. Get Access Token	## Google Cloud Auth Reads Service Account credentials from Google Sheets, signs a JWT locally, and exchanges it for a short-lived OAuth access token. Runs fresh on every request.
3. Get Access Token	HTTP Request	Exchanges JWT assertion for Google OAuth access token	2. Build JWT from Sheet	Check Auth Token Valid	## Google Cloud Auth Reads Service Account credentials from Google Sheets, signs a JWT locally, and exchanges it for a short-lived OAuth access token. Runs fresh on every request.
Check Auth Token Valid	If	Branches based on whether OAuth token retrieval succeeded	3. Get Access Token	Send Processing Message, Send Validation Error	## Google Cloud Auth Reads Service Account credentials from Google Sheets, signs a JWT locally, and exchanges it for a short-lived OAuth access token. Runs fresh on every request.
Send Validation Error	Telegram	Sends an error message to the Telegram user	Check Auth Token Valid
Send Processing Message	Telegram	Notifies the user that video generation has started	Check Auth Token Valid	Download Image	## Download & Convert Downloads the highest-resolution version of the photo from Telegram and converts it to Base64. Any conversion failure is caught and reported back to the user.
Download Image	Telegram	Downloads the Telegram photo file	Send Processing Message	Convert Image to Base64	## Download & Convert Downloads the highest-resolution version of the photo from Telegram and converts it to Base64. Any conversion failure is caught and reported back to the user.
Convert Image to Base64	Code	Converts binary image content into Base64 and MIME metadata	Download Image	Conversion OK?	## Download & Convert Downloads the highest-resolution version of the photo from Telegram and converts it to Base64. Any conversion failure is caught and reported back to the user.
Conversion OK?	If	Branches on image conversion success or failure	Convert Image to Base64	5. Prepare Veo Request, Send Conversion Error	## Download & Convert Downloads the highest-resolution version of the photo from Telegram and converts it to Base64. Any conversion failure is caught and reported back to the user.
Send Conversion Error	Telegram	Sends image-processing failure message to the user	Conversion OK?		## Download & Convert Downloads the highest-resolution version of the photo from Telegram and converts it to Base64. Any conversion failure is caught and reported back to the user.
5. Prepare Veo Request	Code	Builds prompt and JSON payload for Veo 3	Conversion OK?	6. Call Vertex AI Veo 3	## Submit to Veo 3 Builds the API payload with a cinematic 360° orbit prompt and submits a long-running job to Vertex AI Veo 3. Gets back an operation ID used for polling.
6. Call Vertex AI Veo 3	HTTP Request	Submits long-running Veo generation request to Vertex AI	5. Prepare Veo Request	Extract Operation Name	## Submit to Veo 3 Builds the API payload with a cinematic 360° orbit prompt and submits a long-running job to Vertex AI Veo 3. Gets back an operation ID used for polling.
Extract Operation Name	Code	Extracts polling operation name from Vertex AI response	6. Call Vertex AI Veo 3	Wait 2 Minutes	## Poll for Result Checks every 2 minutes whether the video is ready. Times out after 40 attempts (~10 min) and notifies the user with a friendly error if generation takes too long.
Wait 2 Minutes	Wait	Delays the next polling attempt by 2 minutes	Extract Operation Name, Check Timeout	Poll Video Status	## Poll for Result Checks every 2 minutes whether the video is ready. Times out after 40 attempts (~10 min) and notifies the user with a friendly error if generation takes too long.
Poll Video Status	HTTP Request	Polls Vertex AI for long-running operation status	Wait 2 Minutes	Is Video Ready?	## Poll for Result Checks every 2 minutes whether the video is ready. Times out after 40 attempts (~10 min) and notifies the user with a friendly error if generation takes too long.
Is Video Ready?	If	Checks whether the operation has completed	Poll Video Status	Convert Video to File, Continue Polling	## Poll for Result Checks every 2 minutes whether the video is ready. Times out after 40 attempts (~10 min) and notifies the user with a friendly error if generation takes too long.
Continue Polling	Code	Increments poll count and emits timeout or next polling state	Is Video Ready?	Check Timeout	## Poll for Result Checks every 2 minutes whether the video is ready. Times out after 40 attempts (~10 min) and notifies the user with a friendly error if generation takes too long.
Check Timeout	If	Branches between timeout handling and another polling cycle	Continue Polling	Send Timeout Error, Wait 2 Minutes	## Poll for Result Checks every 2 minutes whether the video is ready. Times out after 40 attempts (~10 min) and notifies the user with a friendly error if generation takes too long.
Send Timeout Error	Telegram	Informs the user that generation timed out	Check Timeout		## Poll for Result Checks every 2 minutes whether the video is ready. Times out after 40 attempts (~10 min) and notifies the user with a friendly error if generation takes too long.
Convert Video to File	Convert to File	Converts returned Base64 video into binary file data	Is Video Ready?	Send Video to User	## Deliver Video Converts the Base64 video bytes to a file and sends it directly to the user in Telegram.
Send Video to User	Telegram	Sends the generated video back to the Telegram user	Convert Video to File		## Deliver Video Converts the Base64 video bytes to a file and sends it directly to the user in Telegram.
📋 Overview	Sticky Note	Workspace documentation and setup instructions			## 🎬 360° Product Video Generator Turn a single product photo into a cinematic 360° video using Google Veo 3 — delivered straight to Telegram. ## How it works 1. User sends a product photo to your Telegram bot 2. The workflow authenticates with Google Cloud using a Service Account stored in Google Sheets 3. The image is sent to Vertex AI (Veo 3) with a 360° orbit camera prompt 4. The workflow polls every 2 minutes until the video is ready (up to 10 min) 5. The finished video is sent back to the user in Telegram ## Setup steps 1. Create a Telegram bot via @BotFather and add the credentials in n8n 2. Enable the Vertex AI API in your Google Cloud project and request Veo 3 preview access 3. Create a Service Account with `roles/aiplatform.user` and download the JSON key 4. Paste the key fields into your Google Sheet — columns needed: `client_email`, `private_key`, `project_id`, `scope` 5. Update the 1. Get Service Account Details node with your Sheet ID 6. Connect your Google and Telegram credentials, then activate the workflow
Group 1	Sticky Note	Visual group note for input validation block			## Receive & Validate Listens for Telegram messages and checks that a photo was sent and is at least 480px. Rejects documents or text-only messages and replies with a clear error.
Group 2	Sticky Note	Visual group note for auth block			## Google Cloud Auth Reads Service Account credentials from Google Sheets, signs a JWT locally, and exchanges it for a short-lived OAuth access token. Runs fresh on every request.
Group 3	Sticky Note	Visual group note for image processing block			## Download & Convert Downloads the highest-resolution version of the photo from Telegram and converts it to Base64. Any conversion failure is caught and reported back to the user.
Group 4	Sticky Note	Visual group note for Veo submission block			## Submit to Veo 3 Builds the API payload with a cinematic 360° orbit prompt and submits a long-running job to Vertex AI Veo 3. Gets back an operation ID used for polling.
Group 5	Sticky Note	Visual group note for polling block			## Poll for Result Checks every 2 minutes whether the video is ready. Times out after 40 attempts (~10 min) and notifies the user with a friendly error if generation takes too long.
Group 6	Sticky Note	Visual group note for delivery block			## Deliver Video Converts the Base64 video bytes to a file and sends it directly to the user in Telegram.

4. Reproducing the Workflow from Scratch

Create a new workflow in n8n and name it something like:
Generate 360° product videos from photos with Veo 3 and Telegram.
Create a Telegram Trigger node
- Node type: Telegram Trigger
- Configure Telegram credentials using a bot created via @BotFather
- Set update type to message
- This is the main entry point
Add a Code node named 2. Validate Input
- Connect it after Telegram Trigger
- Use JavaScript to:
  - verify message exists
  - verify message.photo exists and is a non-empty array
  - select the largest photo variant
  - reject images smaller than 480 px on the shortest side
  - extract:
    - chatId
    - messageId
    - caption
    - file_id
  - add error and errorMessage fields when validation fails
- Use the same validation logic as in the workflow JSON
Add a Google Sheets node named 1. Get Service Account Details
- Connect it after 2. Validate Input
- Configure Google Sheets OAuth2 credentials
- Select the spreadsheet containing service account fields
- Select the target sheet/tab
- Ensure the sheet contains columns:
  - client_email
  - private_key
  - project_id
  - scope
- Replace the placeholder sheet ID with your real Google Sheet ID
Add a Code node named 2. Build JWT from Sheet
- Connect it after 1. Get Service Account Details
- In JavaScript:
  - import crypto
  - read client_email, private_key, project_id, scope
  - replace escaped \n in the private key with actual newlines
  - build JWT header and claims
  - sign with RSA-SHA256
  - output:
    - jwt
    - project_id
    - client_email
    - error
    - errorMessage on failure
Add an HTTP Request node named 3. Get Access Token
- Connect it after 2. Build JWT from Sheet
- Method: POST
- URL: https://oauth2.googleapis.com/token
- Content type: Form URL Encoded
- Add body parameters:
  - grant_type = urn:ietf:params:oauth:grant-type:jwt-bearer
  - assertion = {{$json.jwt}}
Add an If node named Check Auth Token Valid
- Connect it after 3. Get Access Token
- Condition should verify that access_token is not empty
- True branch = continue
- False branch = send error
Add a Telegram node named Send Validation Error
- Connect it to the false output of Check Auth Token Valid
- Operation: send message
- Chat ID: {{$json.chatId}}
- Text: {{$json.errorMessage}}
- Set reply-to message ID to {{$json.messageId}}
Add a Telegram node named Send Processing Message
- Connect it to the true output of Check Auth Token Valid
- Operation: send message
- Chat ID: {{$json.chatId}}
- Text should indicate generation has started, for example:
  - creating the 360° video
  - estimated wait time
  - processing with Veo 3
- Set reply-to message ID to {{$json.messageId}}
Add a Telegram node named Download Image
- Connect it after Send Processing Message
- Resource: file
- Provide a file ID expression
- The current workflow uses: {{$json.result.reply_to_message.photo[3].file_id}}
- For a more robust rebuild, prefer carrying forward and using: {{$json.file_id}}
- Ensure binary download is enabled as needed by your n8n version
Add a Code node named Convert Image to Base64
- Connect it after Download Image
- Implement logic to:
  - detect binary payload shape
  - convert image bytes to Base64
  - capture MIME type
  - return conversionStatus = success
  - on error return conversionStatus = failed and conversionError
Add an If node named Conversion OK?
- Connect it after Convert Image to Base64
- Condition:
  - conversionStatus equals success
- True branch = continue to Veo request
- False branch = send conversion error
Add a Telegram node named Send Conversion Error
- Connect it to the false output of Conversion OK?
- Operation: send message
- Chat ID: {{$json.chatId}}
- Text should include {{$json.conversionError}}
- Set reply-to message ID to {{$json.messageId}}
Add a Code node named 5. Prepare Veo Request
- Connect it to the true output of Conversion OK?
- Build a prompt describing:
  - a professional 360-degree product showcase
  - smooth orbit motion
  - white background
  - centered product
  - cinematic movement
- Append Telegram caption as additional context if present
- Build payload:
  - instances[0].prompt
  - instances[0].image.bytesBase64Encoded
  - instances[0].image.mimeType
  - parameters.aspectRatio = 16:9
  - parameters.sampleCount = 1
  - parameters.durationSeconds = 8
  - parameters.personGeneration = allow_all
  - parameters.addWatermark = true
  - parameters.includeRaiReason = true
  - parameters.generateAudio = true
- Return:
  - serialized JSON payload
  - chatId
  - messageId
  - caption
Add an HTTP Request node named 6. Call Vertex AI Veo 3
- Connect it after 5. Prepare Veo Request
- Method: POST
- URL: https://us-central1-aiplatform.googleapis.com/v1/projects/{{$('3. Get Access Token').item.json.project_id}}/locations/us-central1/publishers/google/models/veo-3.0-generate-preview:predictLongRunning
- Send JSON body using the prepared payload
- Add headers:
  - Authorization = Bearer {{$('3. Get Access Token').item.json.access_token}}
  - Content-Type = application/json
Add a Code node named Extract Operation Name
- Connect it after 6. Call Vertex AI Veo 3
- Read the API response
- If name exists:
  - set operation_name
  - set status = polling
  - set poll_count = 0
  - carry chatId, messageId, caption
- Otherwise mark the item as failed
Add a Wait node named Wait 2 Minutes
- Connect it after Extract Operation Name
- Wait time: 2 minutes
Add an HTTP Request node named Poll Video Status
- Connect it after Wait 2 Minutes
- Method: POST
- URL: https://us-central1-aiplatform.googleapis.com/v1/projects/{{$('3. Get Access Token').item.json.project_id}}/locations/us-central1/publishers/google/models/veo-3.0-generate-preview:fetchPredictOperation
- JSON body:
  - operationName = {{$json.operation_name}}
- Add the same Authorization header as the submit node if required:
  - Authorization = Bearer {{$('3. Get Access Token').item.json.access_token}}
  - Content-Type = application/json
Add an If node named Is Video Ready?
- Connect it after Poll Video Status
- Condition:
  - done equals true
- True branch = convert and send video
- False branch = continue polling
Add a Code node named Continue Polling
- Connect it to the false output of Is Video Ready?
- Increment poll_count
- If count exceeds 40, output:
  - status = timeout
  - chatId
  - messageId
- Otherwise output:
  - operation_name
  - poll_count
  - status = polling
  - chatId
  - messageId
  - caption
Add an If node named Check Timeout
- Connect it after Continue Polling
- Condition:
  - status equals timeout
- True branch = send timeout error
- False branch = loop back to Wait 2 Minutes
Connect the false output of Check Timeout back to Wait 2 Minutes
- This forms the polling loop.
Add a Telegram node named Send Timeout Error
- Connect it to the true output of Check Timeout
- Operation: send message
- Chat ID: {{$json.chatId}}
- Text should mention timeout and include {{$json.poll_count}}
- Set reply-to message ID to {{$json.messageId}}
Add a Convert to File node named Convert Video to File
- Connect it to the true output of Is Video Ready?
- Operation: toBinary
- Source property: response.videos[0].bytesBase64Encoded
Add a Telegram node named Send Video to User
- Connect it after Convert Video to File
- Operation: sendVideo
- Enable binary data sending
- Current workflow uses chat ID from: {{$('Telegram Trigger').item.json.message.chat.id}}
- For a more robust rebuild, prefer using: {{$json.chatId}} if you preserve it through the polling chain
Add workspace sticky notes if desired
- One overview note with setup requirements
- Group notes for:
  - Receive & Validate
  - Google Cloud Auth
  - Download & Convert
  - Submit to Veo 3
  - Poll for Result
  - Deliver Video
Configure required credentials
- Telegram API credentials
  - Connect the same bot credentials to:
    - Telegram Trigger
    - Send Validation Error
    - Send Processing Message
    - Download Image
    - Send Conversion Error
    - Send Timeout Error
    - Send Video to User
- Google Sheets OAuth2 credentials
  - Connect them to 1. Get Service Account Details
Prepare Google Cloud outside n8n
- Enable Vertex AI API
- Request access to Veo 3 preview if required
- Create a Service Account with role:
  - roles/aiplatform.user
- Download the JSON key
- Put these values into Google Sheets:
  - client_email
  - private_key
  - project_id
  - scope
Test with a Telegram photo
- Send a photo, not a document
- Prefer image sizes above 1024×1024 for better results
- Verify:
  - processing message is sent
  - image downloads correctly
  - Vertex AI accepts the prompt
  - polling works
  - final video is sent
Recommended hardening before production
- Branch invalid input immediately after 2. Validate Input
- Preserve chatId, messageId, caption, and file_id across Google Sheets and HTTP nodes
- Use file_id directly in Download Image
- Add Authorization headers to Poll Video Status
- Add explicit error handling for Vertex AI request failures
- Use {{$json.chatId}} for final delivery rather than referencing the trigger node

5. General Notes & Resources

Note Content	Context or Link
Turn a single product photo into a cinematic 360° video using Google Veo 3 — delivered straight to Telegram.	Workflow overview
Create a Telegram bot via @BotFather and add the credentials in n8n.	Telegram setup
Enable the Vertex AI API in your Google Cloud project and request Veo 3 preview access.	Google Cloud setup
Create a Service Account with `roles/aiplatform.user` and download the JSON key.	IAM setup
Paste the key fields into your Google Sheet — columns needed: `client_email`, `private_key`, `project_id`, `scope`.	Credential storage design
Update the `1. Get Service Account Details` node with your Sheet ID.	Google Sheets node setup
Connect your Google and Telegram credentials, then activate the workflow.	Deployment step

46 KiB Raw Blame History Unescape Escape

Generate 360° product videos from photos with Veo 3 and Telegram

1. Workflow Overview

1.1 Input Reception and Validation

1.2 Google Cloud Authentication

1.3 Image Download and Conversion

1.4 Veo 3 Job Submission

1.5 Polling and Timeout Handling

1.6 Video Delivery

2. Block-by-Block Analysis

2.1 Input Reception and Validation

Node: Telegram Trigger

Node: 2. Validate Input

2.2 Google Cloud Authentication

Node: 1. Get Service Account Details

Node: 2. Build JWT from Sheet

Node: 3. Get Access Token

Node: Check Auth Token Valid

Node: Send Validation Error

Node: Send Processing Message

2.3 Image Download and Conversion

Node: Download Image

Node: Convert Image to Base64

Node: Conversion OK?

Node: Send Conversion Error

2.4 Veo 3 Job Submission

Node: 5. Prepare Veo Request

Node: 6. Call Vertex AI Veo 3

Node: Extract Operation Name

2.5 Polling and Timeout Handling

Node: Wait 2 Minutes

Node: Poll Video Status

Node: Is Video Ready?

Node: Continue Polling

Node: Check Timeout

Node: Send Timeout Error

2.6 Video Delivery

Node: Convert Video to File

Node: Send Video to User

3. Summary Table

4. Reproducing the Workflow from Scratch

5. General Notes & Resources

46 KiB

Raw Blame History