Turn images in Google Drive into clear written notes in Google Docs. Great for teams that need fast, consistent image summaries for listings, audits, or reports. Runs local vision models so content stays on your machine.

Here is how it works. You start the run manually. The workflow pulls a Google Drive file using its file ID, converts the image to base64, and builds a list of vision models to test. It then loops through each model, sends a structured prompt and the image to your local model server, collects the results, and writes the descriptions into a Google Doc. Two prompt presets are included, one general and one for spreadsheet images, so you can pick the style you need.

To set it up, you need Google Drive and Google Docs credentials in n8n and a local model server running at the default address. Add your Google Drive image file ID and your Google Doc ID once, then run a test to confirm it writes the output. Expect faster documentation for real estate listings, marketing briefs, and research notes. Most teams see big time savings when comparing multiple models in one pass.

• Manual start for safe test runs and controlled execution
• Google Drive download of a specific image by file ID
• Base64 conversion of the downloaded image for model input
• Preset list of vision models with an easy way to edit the list
• Looping over each model using split and batch nodes to manage throughput
• HTTP request to a local model endpoint with a structured JSON body
• Two prompt presets for general images and spreadsheet screenshots
• Creates a tidy result object and saves responses to Google Docs
• Markdown styled descriptions for readable output in documents

• Reduce manual image description work from 30 minutes to 3 minutes per file
• Streamline multi model testing by running all models in one pass
• Improve consistency of image notes by using a fixed prompt template
• Keep data private by processing images on a local model server
• Connect Google Drive and Google Docs to move results without copy paste

1. Import the template into n8n: Create a new workflow in n8n > Click the three dots menu > Select 'Import from File' > Choose the downloaded JSON file.
2. You'll need accounts with Google Drive, Google Docs and Ollama. See the Tools Required section above for links to create accounts with these services.
3. Install and run your local model server so it listens at http://127.0.0.1:11434. Pull the vision models you plan to use such as granite3.2-vision, llama3.2-vision, or gemma3.
4. In n8n, open the Google Drive node. In the Credential to connect with field, click Create new credential and follow the on screen steps to add Google Drive OAuth2. Use a name you will recognize later.
5. Open the Google Docs node. Create a new Google Docs OAuth2 credential the same way and save it.
6. Open the Set node labeled Google Doc Image Id. Add a field named id and paste your Google Drive image file ID. If your Google Docs node requires a document ID field in its parameters, paste that Doc ID there as well.
7. Open the List of Vision Models node and edit the array to include the models you want to compare.
8. Select your prompt style. Keep General Image Prompt enabled for normal images or enable the Real Estate Spreadsheet Prompt for table or sheet screenshots.
9. Click Test workflow. Confirm the image downloads, the base64 field appears, and each model returns a response. Check the Google Doc to see the new descriptions added.
10. If the HTTP request fails, make sure the local server is running and that the models are pulled. If the Drive download fails, check file permissions and that the file ID is correct. If the Doc is not updated, verify the Doc ID and that your credential has edit access.