Description Generator is a free, open-source tool that uses Meta’s Llama model to create product descriptions from images you upload.
This tool supports multiple languages and works by analyzing the image and generating relevant, promotional descriptions.
You upload an image and the AI transforms the product image into compelling sales copy.
For example, a fireworks image input produced this output:
"This Fireworks 4-Pack offers vibrant colors and patterns for any celebration. Featuring sparkling bursts and crackling whistles, it's ideal for special occasions or backyard gatherings. Light up the night with this exciting set."
The results show that the descriptions are focused on highlighting the product’s features and value. This can be useful in content creation for e-commerce platforms, marketing materials, and product listings across various markets.
How to use it:
1. To use Description Generator, you have two options: either deploy it locally or use the demo on Vercel. Below are the steps to help you get started.
2. Deploy Locally:
- Clone the repository:
git clone https://github.com/Nutlope/description-generator - Create a
.envfile and add your Together AI API key:
TOGETHER_API_KEY=YOUR_API_KEY- Set up an S3 bucket and add credentials to the
.envfile using the following values:
S3_UPLOAD_KEY=YOUR_KEY
S3_UPLOAD_SECRET=YOUR_SECRET
S3_UPLOAD_BUCKET=YOUR_BUCKET
S3_UPLOAD_REGION=YOUR_REGION- Install dependencies and run the app locally:
npm install
npm run dev
3. You can also try the demo app hosted on Vercel.
4. Upload an image of your product for description generation.
5. Select the Llama vision model you want to use: Llama 3.2 11B or Llama 3.2 90B
6. Choose up to 3 languages for the product descriptions.
- English
- Spanish
- French
- German
- Italian
- Japanese
- Korean
- Chinese
- Portuguese
7. Select the length of the product descriptions.
- Short
- Medium
- Long
8. Hit the “Generate descriptions” button to create product descriptions.
The live demo might occasionally encounter errors due to server limitations. For better performance, deploying the tool locally is recommended.
