🎬 Automate video uploads to thumbnails with FFmpeg and Google Drive

⚡ 1,476 views · 🎬 Content Creation & Video

Description

Automate Video Upload → Auto-Thumbnail → Google Drive

This workflow accepts a video via HTTP upload, verifies it’s a valid video, extracts a thumbnail frame at the 5-second mark using FFmpeg (auto-installs a static build if missing), uploads the image to a specified Google Drive folder and returns a structured JSON response containing the new file’s details.

Who’s it for

Marketing / Social teams who need ready-to-publish thumbnails from raw uploads.
Developers who want an API-first thumbnail microservice without standing up extra infrastructure.
Agencies / Creators standardizing assets in a shared Drive.

How it works

Accept Video Upload (Webhook) Receives multipart/form-data with file in field media at /mediaUpload. Response is deferred until the final node.
Validate Upload is Video (IF) Checks {{$binary.media.mimeType}} contains video/. Non-video payloads can be rejected with HTTP 400.
Persist Upload to /tmp (Write Binary File) Writes the uploaded file to /tmp/<originalFilename or input.mp4> for stable processing.
Extract Thumbnail with FFmpeg (Execute Command)
- Uses system ffmpeg if available; otherwise downloads a static binary to /tmp/ffmpeg.
- Runs: ffmpeg -y -ss 5 -i <video> -frames:v 1 -q:v 2 /tmp/thumbnail.jpg
Load Thumbnail from Disk (Read Binary File) Reads /tmp/thumbnail.jpg into the item’s binary as thumbnail.
Upload Thumbnail to Drive (Google Drive) Uploads to your target folder. File name is <original>-thumb.jpg.
Return API Response (Respond to Webhook) Sends JSON to the client including Drive file id, name, links, size, and checksums (if available).

How to set up

Import the workflow JSON into n8n.
Google Drive
- Create (or choose) a destination folder; copy its Folder ID.
- Add Google Drive OAuth2 credentials in n8n and select them in the Drive node.
- Set the folder in the Drive “Upload” node.

Webhook

The endpoint is POST /webhook/mediaUpload.

Test:

curl -X POST https://YOUR-N8N-URL/webhook/mediaUpload \
  -F "media=@/path/to/video.mp4"

FFmpeg
- Nothing to install manually: the Execute Command node auto-installs a static ffmpeg if it’s not present.
- (Optional) If running n8n in Docker and you want permanence, use an image that includes ffmpeg.
Response body
- The Respond node returns JSON with file metadata. You can customize the fields as needed.
(Optional) Non-video branch
- On the IF node’s false output, add a Respond node with HTTP 400 and a helpful message.

Requirements

n8n instance with Execute Command node enabled (self-hosted/container/VM).
Outbound network access (to fetch static FFmpeg if not installed).
Google Drive OAuth2 credential with permission to the destination folder.
Adequate temp space in /tmp for the uploaded video and generated thumbnail.

How to customize

Timestamp: change -ss 5 to another second, or parameterize it via query/body (e.g., timestamp=15).
Multiple thumbnails: duplicate the FFmpeg + Read steps with -ss 5, -ss 15, -ss 30, suffix names -thumb-5.jpg, etc.
File naming: include the upload time or Drive file ID: {{ base + '-' + $now + '-thumb.jpg' }}.
Public sharing: add a Drive → Permission: Create node (Role: reader, Type: anyone) and return webViewLink.
Output target: replace the Drive node with S3 Upload or Zoho WorkDrive (HTTP Request) if needed.
Validation: enforce max file size/MIME whitelist in a small Function node before writing to disk.
Logging: append a row to Google Sheets/Notion with sourceFile, thumbId, size, duration, status.

Add-ons

Slack / Teams notification with the uploaded thumbnail link.
Image optimization (e.g., convert to WebP or resize variants).
Retry & alerts via error trigger workflow.
Audit log to a database (e.g., Postgres) for observability.

Use Case Examples

CMS ingestion: Editors upload videos; workflow returns a thumbnail URL to store alongside the article.
Social scheduling: Upload longform to generate a quick hero image for a post.
Client portals: Clients drop raw footage; you keep thumbnails uniform in one Drive folder.

Common troubleshooting

Issue	Possible Cause	Solution
`ffmpeg: not found`	System lacks ffmpeg and static build couldn’t download	Ensure outbound HTTPS allowed; keep the auto-installer lines intact; or use a Docker image that includes ffmpeg.
Webhook returns 400 “not a video”	Wrong field name or non-video MIME	Send file in `media` field; ensure it’s `video/*`.
Drive upload fails (`403` / insufficient permissions)	OAuth scope or account lacks folder access	Reconnect Drive credential; verify the destination Folder ID and sharing/ownership.
Response missing `webViewLink` / `webContentLink`	Drive node not returning link fields	Enable link fields in the Drive node or build URLs using the returned `id`.
`413 Payload Too Large` at reverse proxy	Proxy limits on upload size	Increase body size limits in your proxy (e.g., Nginx `client_max_body_size`).
Disk full / ENOSPC	Large uploads filling `/tmp`	Increase temp storage; keep Cleanup step; consider size caps and early rejection.
Corrupt thumbnail or black frame	Timestamp lands on a black frame	Change `-ss` or use `-ss` before `-i` vs. after; try different seconds (e.g., 1–3s).
Slow extraction	Large or remote files; cold FFmpeg download	Warm the container; host near upload source; keep static ffmpeg cached in image.
Duplicate outputs	Repeat requests with same video/name	Add a de-dup check (query Drive for existing `<base>-thumb.jpg` before upload).

Need Help?

Want this wired to S3 or Zoho WorkDrive or to generate multiple timestamps and public links out of the box? We’re happy to help.

🔗 Nodes Used

Read Binary File, Write Binary File, Webhook, Google Drive

📥 Import

Download workflow.json and import into n8n: Workflow menu → Import from File

📖 Importing guide · 🔑 Credential setup