DALL-E Mini: All about AI that generates images

Tiempo de lectura: 3 minutos

Undoubtedly, the year 2022 has left us with some unusual images, has brought unexpected changes and has shown us that technology is taking giant steps forward. If we talk about unusual images, we cannot overlook those generated by the artificial intelligence that has given so much to talk about: DALL-E 2, Google Image, DALL-E Mini, etc.

Today we want to bring you closer to the reality that artificial intelligence has become, and for this, we want to talk about the curious tool DALL-E Mini. This open-source artificial intelligence has gained the attention of the general public for its curious abilities. Don鈥檛 you know it? Well, read on and we鈥檒l tell you all about it.

馃鈥娐 What is DALL-E Mini?

For those people who have not been aware of the latest developments in the artificial intelligence sector, the names OpenAI and DALL-E will sound completely unfamiliar. And probably, the name DALL-E Mini (or Craiyon), will be an even bigger mystery.

DALL-E Mini (as its name suggests) is the 鈥漧ittle sister鈥 of DALL-E 2. Like DALL-E 2, is an artificial intelligence capable of generating images in seconds, just by giving it a few simple commands. However, unlike its big sister, DALL-E Mini has not been created by OpenAI, the company dedicated to artificial intelligence research led by Elon Musk.

DALL E Mini is an open-source artificial intelligence, developed by programmer Boris Dayma. You could say that it is a version with less power and fewer resources, but it is still capable of generating amazing results.

Do you want to learn about Artificial Intelligence? 馃 We make it easy for you with our Data Science course

馃柤锔徛 How does DALL-E Mini work?

Now that we鈥檝e made its origin clear, you鈥檙e probably wondering where you can start using it and how DALL-E Mini works.聽

The answer is simple: just access the Craiyon page (this is the new name of DALL-E Mini) and use its menu to write what you want the artificial intelligence to generate for you in the form of an image.聽

men煤 de dall e mini

Although you can make your requests in Spanish, we have done some tests and it seems to offer more accurate results if your request is written in English. That鈥檚 why we recommend that you try it in both languages and compare.

The results are generated by collecting images from the Internet and end up creating a collage with 9 images that try to fulfill the request you have made. Here are some of the results:

solicitud de dibujo a dall e mini

The truth is that this drawing AI has a surprising capacity, considering its limitations. Although it doesn鈥檛 quite get what we asked it to do right (to generate images of Albert Einstein sightseeing in Barcelona), it gives us results quickly.聽

Craiyon has earned affection among users, since its results are somewhat bizarre, something that has made people have fun trying different and innovative requests.

Still, we should not forget that all artificial intelligence needs a lot of testing to improve its results. That is why every time someone makes a request, DALL-E Mini improves progressively, offering more and more accurate results.

How is DALL-E Mini different from other artificial intelligence?

As we have pointed out throughout the post, different types of artificial intelligence generate images, but not all of them are the same. Among the most talked about are DALL-E 2, Imagen and Drawanyone.

We have already talked about the first one, but we cannot forget about Imagen: an artificial intelligence created by the Google Research team. This AI has become the direct competitor of the tool created by OpenAI. The reason is none other than its incredible ability to create highly realistic images.聽

ejemplos de im谩genes generadas por google imagen

Even so, the Image is currently only available to a limited group of people. This undoubtedly opens a debate about the possibilities of this artificial intelligence, and whether their use is open to everyone or not.

In the case of Drawanyone, we find a very interesting tool, which generates very accurate drawings of people by providing just 10 photos of a face. As in the case of Image, Drawanyone also has limitations, as the time it takes to generate these drawings is much longer (about an hour).

Undoubtedly, artificial intelligence has come to stay in our lives and offers us hitherto inconceivable possibilities. Whether for generating texts, images or programming code, we are facing a total paradigm shift. If you don鈥檛 want to miss this train, we recommend that you continue reading about all that artificial intelligence has to offer, such as the amazing ChatGPT.

Carlos Palm贸n
Carlos Palm贸n es soci贸logo, consultor SEO y copywriter. Apasionado por entender la sociedad, se lanz贸 a estudiar el grado en Sociolog铆a. En 2020, descubri贸 en el marketing digital una forma de aunar su inter茅s sociol贸gico con sus aptitudes creativas.