Before ChatGPT, Artificial Intelligence platforms arrived that allow us to create images based on text. Generating images from text using AI is much more complicated than some users may initially think, as long as they want to get the best results at all times. While it is true that, with a simple line, we can obtain spectacular images, if we really want to obtain what we want, we must specify as much data as possible.
Midjourney and Stable Diffusion are the most well-known and used Artificial Intelligence platforms, however, they are not the only ones that allow us to generate images out of nothing, based on a vague or complete description such as Lexica, one of the best options for create drawings or anime characters. If we want to start delving into the world of image generation using Artificial Intelligence, as long as we want to obtain spectacular images, we must take into account the following sections in the description.
The first thing to take into account is that, given the same description, we will always get different resultswe will never obtain exactly the same image based on the same description, so if we omit any detail, we will have to resort to Photoshop or alternatives, if we like the initial result we have obtained and do not want to discard it.
- Background. A section that very few people pay attention to when generating images using AI is the background. The background, in any photograph, provides a context, a context that allows us to situate the action. The same happens in videos, which is why the number of detractors of vertical videos is so high, as well as for obvious reasons (monitors and televisions do not rotate like mobile phones).
- Atmosphere. If we like a specific theme, a painter, a specific genre or a very specific aesthetic, we can add this influence to the description of the image we want to create to obtain an image that includes those touches that we have always liked.
- Design. When creating anime characters, we must take into account the design of the clothes that we want them to wear, as well as the accessories that we want them to wear. Yes, we want it to show a katana and that it includes certain details in the handle, we must specify it in the description, where we can also establish what pose the character has.
- particularities. To obtain the best results, we must take into account all the details, including some particularities such as, for example, having a scar on the right eye, having a flat nose, a small mouth, curly hair… details that, at first glance, It seems insignificant but it can offer us more than spectacular and unique results.
It is possible that some of the details in the description are bypassed by AIThis is something that we cannot avoid, especially when we talk about very specific features, so we do not get the results we are looking for, we will have to resort to Photoshop to add them.