As I am from Asia, especially Japan, I let the AI draw a photo of me in traditional kimono, kimono, yukata and hakama, with Kyoto as the background setting. The results were very disappointing, with a mixture of Asian and Japanese images, ornaments, clothing styles, patterns used in traditional costumes and colour schemes that did not look Japanese at all. I also tried adding Kyudo as another keyword, but this also did not show any bows used in Japanese archery. Nor any Japanese umbrellas or other accessories. The facial features and eyes are also different, and there is no apology, profundity or modest smile, which Japan has traditionally regarded as virtues in the past. It became worse when key words such as era were added to determine the historical background. It would be great if the AI image generation engine could be improved to focus on these points.
Translated with DeepL.com (free version)
1
Reply
AdChoices