You might be familiar with this situation: you open a Google Image search, looking for something specific. But for some reason, Google decides to show you the best-ranked image, which isn’t always the most suitable one for your search. This can become even more challenging the more precise your search terms become.
In most cases, the ‘best’ image isn’t necessarily the first one you see. So, you end up scrolling and sifting through countless options to find what you’re really looking for. The same challenge applies to many other image sources like Unsplash.
We use AI for so many tasks these days, so why not leverage it to save us from this tedious search for the perfect image? Imagine simply telling AI what kind of image to look for.
In this blog post, I will show you how you can do exactly this using a simple, free tool or by integrating an API into your existing workflow.
Getting Started
The application supports multiple image sources, including URLs, local files, or Unsplash. If you plan to use Unsplash, you’ll need a free API key, which you can obtain here:
- https://unsplash.com/documentation#creating-a-developer-account
- https://unsplash.com/oauth/applications
For all other options, you only need an OpenAI API key, which is described here:
You can try the application here or on RapidAPI using the free plan:
But let’s see how it works first.
Example: Unsplash
Unsplash is one of the most popular image platforms out there. While you can search for images there, the perfect image for your specific needs isn’t always the perfect image for everyone else, or for Unsplash’s own ranking algorithm.
In the AI Image Selector, you can add a “choice hint,” such as “scenic image of eiffel tower.”
And just like that, it will choose the image for you:
It doesn’t just select an image for you; it also justifies its choice, explaining why it picked that particular image:
Google Image Search
Let’s say you have a bunch of Google Image results for the Eiffel Tower, like the ones below. The first image is quite simple, while the second and third look a bit more exciting.
To use these, choose the “Custom Image” tab in the app and enter their URLs.
Important: Make sure to open each image individually and then copy its direct URL. Simply copying the image URL from the search results page usually won’t work.
Then, add a custom prompt like “choose best.”
As you can see, the AI did not choose the first image. Instead, it picked one of the more visually appealing options, specifically one with a scenic view.
As mentioned earlier, you can also use this functionality via an API request, which is shown at the bottom of the application page.
You can now embed this API into any workflow. For example, you could use it when creating blog posts to let AI select the images from any source, whether it’s Unsplash, your own files, or even AI-generated images.
Potential Use Cases
- For Content Creators: Automatically choose the best featured image for your blog posts, articles, or social media updates from a selection of options, ensuring they align perfectly with your content’s theme.
- For E-commerce Platforms: Efficiently select the most appealing product photo from a series of shots for online listings, helping to improve visual impact and potentially increase sales.
- For Developers Integrating AI-Generated Content: If your application generates multiple images (e.g., using Midjourney or DALL-E), use the API to programmatically pick the single best image that fits specific criteria, streamlining your content pipeline.
Happy imaging!
Top comments (0)