New artificial intelligence DALL-E

DALL-E

It is a new artificial intelligence that tries to create drawings or images containing a description that you indicate. It is an OpenAI generator whose code is open and you can find it on GitHub. It has been created by the programmer Boris Dayma, who developed the application as part of a Google and Hugging Face competition in July 2021. The idea behind this AI is to create an algorithm to generate images similar to Google’s DALLE, but with a smaller infrastructure and fewer hardware resources.

It works with simple software as simple as typing something you want to see whether it’s an animal, food, a famous person, cartoons or objects. The first step that the software performs is to analyze the words that are in the request that you have written to it. The AI then searches the Internet for those words and looks for the images associated with or related to them, as well as the distinctive features that make that image the definition of the word. What the artificial intelligence proceeds to do with the information you previously provided is to create a composition with fragments of those images that it has searched for, and it tries to make the result as realistic as possible. The images are collected from the net without any kind of filter.

For now, the results created by the software cannot create images or photographs in a hyper-realistic way. All artificial intelligences go through a training or coupling period in which they improve according to feedback from the user, that is, the more people use the software it will have better results. That is the reason why it generates such strange and bizarre images at the beginning and if you are looking for a very natural or real result you will be surprised or perhaps disappointed by the results that it is initially, because it is an algorithm that is still being trained and little by little the results will improve.

Go ahead and use the amazing application, create unique mixes and help the tool to be more and more effective and to perfect the final images, to be able to use it you must access its demo page open to everyone, whose URL address It’s Huggingface.co/spaces/dalle-mini/dalle-mini. On this website, once you are browsing the page you must go to the part where a text box appears and below as a blank space, keep in mind that your searches can only be entered in English, you can make a long and specific description or a mixture of what you want to see, keep in mind that you will not have an exact result or suddenly the software may omit some of the words you wrote, remember that you are using a demo, after writing you must click on the Run button, that’s it You only have to wait and a little and you will see the results that the web has for you, enter and create unique mixtures. It helps the world of technology and artificial intelligence to advance each time and become more effective every day.

On behalf of Google, the artificial intelligence “Image” was created, it has the same functionality as DALL-E, the artificial intelligence developed by Open AI that also allows generating images based on a text description, the difference that “image” has is that Google It has always been characterized by having a great algorithm to identify features faster, “image” achieves better and higher quality results, achieving more realistic results. a text-to-image diffusion model with an unprecedented degree of photorealism and a deep level of language understanding.

Derechos de autor: https://imagen.research.google/

Derechos de autor: https://imagen.research.google/