A blog related to original AI artworks. I make AI artworks using coined words and emoji as prompts.
I considered using my name (Yasusi Kanada) as a prompt. I have written the reasons separately. To achieve various images, I considered specifying only my name as the positive prompt and diversifying the images with negative prompts. Therefore, negative prompts such as Millet, Rembrandt, and others, were effective in drawing pictures resembling abstracted figures when combined with my name (or with “painting”).
First, here are the images drawn by including the name of Millet. Although none of them are highly realistic, various styles of images were obtained, ranging from relatively low to high levels of abstraction, depending on the same positive and negative prompts.
The image on the top right often depicts black individuals when “Yasusi Kanada” is given as a prompt.
The image on the top left appears to depict a cat, but there also seems to be a face overlaid on its nose. The top right image, when a painter’s name is specified, is a relatively uncommon abstract painting, but it also appears to depict a waterfall.
The last two images might be considered Cubism-style landscape paintings, with buildings and something resembling the sun or moon depicted.
Although only six images are presented here, there are both abstract and figurative paintings. In these paintings, prompts such as “Canada,” “Edo,” and “anime” are specified along with Millet, aiming to correct biases in the content of the images. Among the images not shown here, there are even more diverse styles and content.
When Mucha was specified, the following images were drawn.
The image on the left appears to be some kind of automotive part but also resembles a face. The paint seems to be splattered, with the black parts resembling engine oil.
Next, the images drawn by specifying Rembrandt are shown.
The image on the above left contains many objects resembling umbrellas, some of which appear to be turned inside out, although the smaller ones might be mushrooms. In the above right image, three birds are depicted among the trees and flowers, with two of them resembling chickens. There are many other shapes resembling eyes, suggesting the presence of other animals.
Next is Rubens. When Rubens’ spelling is given as a prompt, it often produces more two-dimensional paintings. Rubens, along with Rembrandt, is a representative Dutch painter of the Baroque era and is often compared. Even when the same negative prompt is given, there is a considerable difference in the style of the paintings.
Monet was also tried. Here, along with Monet, “Canada,” “Edo,” and “anime” are specified as prompts, aiming to correct biases in the content of the images.
Arcimboldo was also tried. Although Arcimboldo is not a very famous painter, when the spelling “Arcimboldo” is given to the AI as a prompt, it produces various interesting images, so I tried it here as well.
Various other attempts were made, but I will leave it at this for now. Almost all the images are posted on the website. The images presented here were drawn around October last year. They have recently been uploaded to this site and made viewable. That is the reason why I am writing this article now.