This repo is a NodeJS example of how to upload images and videos to Google's Gemini Vision API. It consists of a simple terminal-based user interface where you're asked if you want to upload a file, select which file, and provide your prompt, and get the response from Gemini.
It's intended to give a simple end-to-end example of making a Gemini API call with files, and is accompanied by a video tutorial of this project.
Prerequisite
- NodeJS installed on your computer
- A Google Cloud account with a valid payment method
Clone the github repo
git clone https://github.com/elfvingralf/gemini-vision-node-example
Navigate to the folder that was created, and then run yarn install to install the dependencies.
If you're not familar with git you can do this instead:
- Download the repo as a zip, or copy/paste the index.js code into a file on your computer
- Open a terminal and run these commands in your folder:
yarn add macos-open-file-dialog, @google-cloud/vertexai, readline
Before you can run the project, we have to set up a Google cloud project for your Gemini API
If you want a visual guide for the steps below, go to 04:40 in this video.
- Go to console.cloud.google.com
- In the top right-hand corner, click the project selector box, then create a new project
- Name your project and proceed (you don't need to select an organization)
- After a few seconds the project should be created, then select it.
- Go back to the project selctor, and copy/paste the project ID. Replace the default value in
const projectId = "REPLACE WITH YOUR GOOGLE CLOUD PROJECT ID"with this id - Exit the project screen and click "APIs and services" in the Quick Access menu
- Select "Enable APIs and Services" on the top bar
- Search for Vertex, which Gemini is a part of. you should see an option "Vertex AI API", click it.
- On the Vertex AI API page, click "Enable" button
Your project now has access to the Vertex AI APIs. But there's one last step before you can run your code locally.
If this is the first time you run Google Cloud APIs on your computer, you need to set up the Google Cloud CLIlocally and initalize your project.
Follow the steps below or use the official documentation.
First download and follow the install instructions in the Google Cloud CLI docs.
When you run ./google-cloud-sdk/bin/gcloud init you should select the project you just created in Google cloud.
Once this is complete, you're ready to run your code
Go to the folder in your terminal. If you haven't already run yarn install do that. Then run:
node index.js
I've included three example files in the project that I use in my video tutorial, but you can of course use any files.
The script will ask if you want to upload a file together with your prompt, and if so ask you to select it. The index.js has a list where you can define which file types are allowed to be selected, as well as a list of file types that are supported by Gemini.
I'll try to add a few more resources to help you extend on this repo/tutorial.
I'm a self-taught and really like scrapping together fun projects. I write functional code that probably isn't beautiful nor efficient, and share it with the hope that someone else might find it useful.
You can find me as @ralfelfving on Twitter/X. If you liked this project, consider checking my tutorials on my YouTube channel @ralfelfving.