This image was pure experimentation. I added the words “Kyoto” and “dragon” to see what I get. I didn’t tell Mijourney AI about the color of the hair, background, or clothes.
The results were quite impressive and super kawaii. A stunning-looking image of a girl wearing Japanese clothing with dragon-shaped art. Her hair color was also matching the orange on the clothes. It all fitted perfectly with the cyan base color of her clothes.
The way her hair merged with the art on the clothes on the right side was just beautiful. I also really like the overall look and the abstract smudgy background.
Like in the previous image of the geisha ballet art, I need to do a lot of work here in Adobe Photoshop.
There are many times, no matter if you write “symmetrical eyes” or try to describe the face, you can get some really weird results. The original crop of the face can be seen below.
You can see the disproportional nose and messy left eye. That needed to be edited in Photoshop. It happened a lot when it comes to a human figures. Maybe I’ll learn a few keywords that will reduce it but until then, I’ll correct it in Photoshop. It doesn’t take too long.
I love this image, especially the colors and the facial expression. The elements of the Dragon are very well apparent as well. I boosted the colors a bit, and brighten up the image but most of it remained untouched.
The anime keyword gave the face a combination of reality with a touch of an anime look. I really love the little smirk and I didn’t want to change it. It’s actually one of the things that really made me want to stay with that. Like in many of my chosen AI art, the facial expression plays a significant role in which image I choose. I want the personality to shine through and talk from the image to the viewer. If I did a good job, you probably understand what I mean.
Oh, I also sharpened the image in Photoshop which really made difference. It brought up the details in the clothing, and the lighting in the lips and eyes and gave the picture a nice elegant painting strokes-like look. Otherwise, it would look too blurry.
TIP: If you use the option with light upscale in Midjourney, it might turn the image to look more realistic as it adds many more details. The based image with “Light Upscale Redo” looks much more anime-cartoony and I wanted to go with something in between. I just felt it’s more interesting that way–at least to what I wanted to achieve with that image.
You can see that it looks more cartoony doll-like. The details in the hair are different.
IMPORTANT INSIGHT: Not just that, each time you upscale the image, it will render it differently. Yes, in another upscale render, the hair was in front of the bandana, the eyes looked different, the lips have slightly different color output, the details in the clothing, and the background, and the way the shirt was rendered beneath the kimono looked very different. It’s not like you are going to get the same results pressing ‘U’ for the same image. Something to keep in mind.
Super cute Kyoto dragon girl portrait AI art. I hope you like the results. More coming, so make sure to visit my blog again soon. Thank you.