[Image Generation Observation Note Day 13] Comparing Japanese and Western culture. How well does AI understand cultural motifs?

Introduction

Image generation AI constructs entire images not only from worldviews like "Japanese-style" or "Western-style," but also from motifs that symbolize culture.

This time, I specified motifs representing Japanese and Western culture to observe how naturally the AI expresses cultural characteristics.

Theme of this session

The comparison theme this time is

Cultural motifs (Japanese culture vs. Western culture)

.

I kept the composition and facial expressions of the characters as fixed as possible, changing only the motifs that represent the culture.

A: Japanese culture

• Dragon

• Sumi-e (ink wash painting)

• Gold leaf

• Cherry blossoms

• Japanese patterns

B: Western culture

• Angel

• Stained glass

• Gothic architecture

• Marble

• Red rose

A: Image of Japanese culture

In the Japanese culture prompt, elements symbolic of Japanese art, such as a gorgeous kimono with a Japanese pattern, dragons, ink wash paintings, gold leaf, landscape paintings, five-story pagodas, and cherry blossoms, were naturally incorporated into the background.

It was impressive that rather than each element existing independently, a sense of unity was created, much like a single Japanese painting.

B: Image of Western culture

In the Western culture prompt, an evening dress was worn, and stained glass, angel statues, angel wings, Gothic architecture, white marble, and red roses were placed in the background.

A space reminiscent of European religious art and palaces was naturally constructed, resulting in a work that felt both glamorous and sacred.

What I learned from the comparison

The most interesting thing about this comparison was that the AI is not just depicting cultural motifs in isolation.

In Japanese culture, not only dragons but also ink wash paintings, gold leaf, and five-story pagodas were naturally combined.

In Western culture as well, not just angels, but stained glass, Gothic architecture, marble, and roses were in harmony, establishing the work as a cohesive piece.

In other words, it appeared that the AI was composing the images while considering the relationships between the elements that symbolize the culture.

What I learned this time

On Day 12, I compared the worldviews of "Japanese style" and "Western style."

On the other hand, on Day 13, I specified more concrete cultural motifs.

As a result, the AI did not just place cultural motifs, but naturally supplemented them with related architecture, decorations, colors, and atmosphere, resulting in an outcome that seemed to integrate them into a single culture.

However, this is an observation result obtained under these specific comparison conditions.

Whether the same trend will hold for other cultures and motifs needs to be verified further in the future.

Summary

In this experiment, the overall atmosphere of the images changed significantly just by using motifs symbolizing Japanese and Western cultures.

It appeared that the AI was not just placing individual motifs side-by-side, but was combining elements associated with each culture to compose the image as a single, harmonious work.

This experiment confirmed that keywords representing culture play an important role in creating a worldview in image generation AI.

Questions for next time

These results have led to new questions.

Even if only one cultural motif is specified, does the AI automatically supplement related cultural elements to construct a worldview?

I would like to observe this point next time.

#ImageGenerationAI
#AIImageGeneration
#ImageGenerationObservationNote
#Prompt
#GenerativeAI
#AIBeginner
#ImageGenerationExperiment
#JapaneseCulture
#WesternCulture
#CulturalMotifs
#JapanesePattern
#Kimono
#GothicArchitecture
#StainedGlass
#AILearning