2022 was the year in which Artificial Intelligence began to reach the general public, a technology that, unlike its first years of operation, advances by leaps and bounds and in a matter of months, changes in terms of the quality of content generated, be it text or images, they have evolved very quickly. If we talk about image generation using Artificial Intelligence, we have to talk about Midjourney.
Midjourney was together with Stable Diffusion, the first two AIs that were released publicly, however, the first one has managed to evolve in a much faster and more efficient way, currently being the best platform to generate images based on a description. Despite being the best option, the developers of this AI are not satisfied and continue working to further improve its operation. With the release of version 5 of Midjourney, one of the most important aspects of generating images has been improved.
Artificial Intelligence platforms are based on a description. While those offer us information via text, such as ChatGPT or Bing Chat, those that allow us to generate images need to know the comma separated dataas if we were describing the object we want to obtain, but without using prepositions.
Natural language support
Version 5 of Midjourney focused on offering higher image quality, a wider range of styles, support for much more realistic textures than the previous version as well as a wide dynamic range and, most importantly, support for using natural language in graphics. prompts. It is no longer necessary to separate the elements that we want to see in the image that we want to generate through this platform, we just have to write with natural language what is the result we are looking for.
Midjourney version 5 is capable of interpret correctly instructions in natural language. But, in addition, it has also introduced changes in the tags that many users used to generate images such as 4K, high resolution, colorful and others, tags that if we use them in the new version of Midjourney, the only thing we are going to achieve is confuse Intelligence Artificial and not get good results.
By introducing support for natural language, these tags have ceased to make sense to the AI and now we can describe the image we want to obtain in text format without being forced to add instructions for it to carry out the process in a certain way. This version invites us to use the same description technique that we learned in school.
If we want to start generating images through Midjourney, the first thing we need to do is join the Discord channel where it is available. In the channel #getting-started All the instructions that we must follow to start generating images on this platform are shown. We can only use English to establish the description of the image that we want to create.
