Stable diffusion, one of the most recent imagers to appear, is constantly growing. Thanks to its open source essence, thousands of people are adapting it to create solutions that we have never seen before, versatility that other generators such as OpenAI’s DALL-E 2 and Meta’s Make-A-Scene do not have due to restrictions on access to their source code.
In a few weeks, the generation model trained with the LAION-Aesthetics image set went from being a complete unknown to one of the most popular. Now that the amount of generated images is growing day by day, some great ones, the developer Sharif Shameem presented Lexicaa image search engine and directions of Stable Diffusion available to everyone.
Searching through millions of images
As we mentioned before, Stable Diffusion has the ability to generate photorealistic images from text, and although the program will respond to anything we ask in English, the results will be more accurate if we use detailed instructions and in an orderly manner. Precisely, the high level of control that users have is one of the advantages of these models.
The problem is that if we use random indications omitting details about the style (do we want a photorealistic image or an oil painting?), the artistic level (we prefer Leonardo DaVinci or Van Gogh), and the lighting parameters (Unreal Engine or triadic scheme ), the result may not be as expected. A resource to “train” us in the use of this AI can be Lexica.
When entering lexica.art we found a very simple interface. Below the name of the search engine, a legend tells us that we will search among more than 5 million images and indicators of Stable Diffusion. Before you start searching you can scroll down the grid and select any of the existing images. This will allow you view indications that have been used.
If you click CopyURL and open the copied link you will be able to find the other compositions generated based on the original indication. In this way you will be able to more easily find inspiration for your own creations and get an idea of the indications that you will have to enter so that they finally come true or are as similar to what you imagine. You can also use the search box on the home page to refine your results.
In case you want to start making your own creations, you have several alternatives to use Stable Diffusion. You can start with DreamStudio, a front-end that works with credits (you get 200 credits when you sign up for free) and is very easy to use. You also have the possibility to combine the capabilities of this AI with Photoshop, GIMP and other tools. There are those who are already doing it and the results are amazing.