It seems that there is a spectrum for artificial intelligence. One side of the spectrum is total wonder and the other extreme obedience. It is easy to be surprised by artificial intelligence; But you can hardly make him obey you completely. If you have a very specific image in mind, it’s like you’re up against a huge force, according to Myro Klingman, who sells NFT AI-generated works. The AI reluctantly obeys commands such as shading a point, enhancing a point, or lowering the shadows of a point. Artificial intelligence should be encouraged to do this.
Current versions of DALL-E, Stable Diffusion, and Midjourney limit commands to about the size of a long tweet. If longer words and sentences are placed next to each other, the image will become more cluttered. In fact, behind every amazing image, there is a magical recipe. It starts with the first magic command. How you say it matters. Instant results appear in sets of four to nine images. You can create the resulting images from the collection of images.
To achieve the ideal image, you need to think like AI: add instructions to your command. Continue with repetition. Change the words to see new results. Then select a few from the created images. Behind this magic is the art of arousal. Every artist or designer has their own way to evoke artificial intelligence. Like directors, these artists guide AI into a unique vision.
PromptBase It’s a marketplace for orderers to sell orders that lead to simple images like logos, icons, avatars, and game weapons. On this website, instead of selling the work of art, you are selling the command or order that leads to the work of art; And unlike fixed clip art, it is easy to change the obtained work based on needs and reach multiple versions of them. Most of these orders are sold at a fair price.
Super-medium commands not only have a subject, but also describe lighting, angle of view, emotion, color palette, degree of abstraction, and maybe even visual reference. For technical reasons, even repeating the same command is unlikely to produce the same output. There is a random kernel for each image without which it is statistically impossible to generate a copy. Furthermore, the same command entered into different AI engines will produce different images. For example, Midjourney images are closer to painting, while DALL-E images resemble real photography. However, the commanders do not want to reveal their secrets.
It seems clear that instructives produce real art. Are film directors like Hitchcock or Kurosawa anything but giving orders to actors, actions, scenes or ideas? Image makers are involved in a similar work, and therefore it is not surprising that they sell their productions in art galleries or even enter them into art competitions. Summer 2022, Joe Allen won first prize in Digital Art at the Colorado State Fair Fine Art Competition for a large canvas with an operatic atmosphere signed by Joe Allen via Midjourney.
Digital art images are usually created using tools like Blender or Photoshop, which allow artists to access libraries of digital objects and textures that are then collaged together to create a landscape. These digital images are actually montages of technology. Collages are a venerable art form, and using artificial intelligence to create collages is a natural evolution. According to Alan, if the collage with 3D rendering is art, then Midjourney’s image is also art.