Chinese Aesthetic Character Prompts by Dynasty (2026)

Tang / Song / Ming / Qing + xianxia / Daoist / guochao / ink-wash: 10 reliable Chinese-aesthetic character prompts, tuned for Midjourney V8.1, Nano Banana 2 and GPT Image 2.

Chinese-aesthetic prompts fail when you write ancient Chinese and stop there. The model has nothing to anchor to, so it mixes Tang sleeves with Ming hairstyles and grades everything in gacha-game gold. Below are 10 dynasty-specific templates plus the five-layer formula (dynasty + material + ornament + environment + art style) that holds the look together. All 10 are copy-ready and tested against the current 2026 image models.

Which model for Chinese-aesthetic portraits (June 2026)

The same prose prompt works across all three of the big generators, but they behave differently on hanfu, embroidery and ink-wash texture. As of June 2026:

ModelStrength for this genreAspect ratioFree tier
Midjourney V8.1Best painterly / cinematic look; native 2K HD output; --oref locks one face across a seriesUse --ar 2:3 etc. in the promptNo free tier; plans from $10/mo
Google Nano Banana 2 (Gemini 3.1 Flash Image)Sharpest skin and fabric micro-texture; true 4K; reads real-world referencesSet in the Gemini app, ignores --arFree users get ~20 images / 24h
GPT Image 2 (ChatGPT Images 2.0)Best prompt adherence for embroidery patterns and text on banners; “Thinking” mode reasons before drawingSet in chat, ignores --arInstant mode free; Thinking mode paid

The templates below are written Midjourney-style with --ar flags. On Nano Banana 2 and GPT Image 2, drop the --ar ... and pick the aspect ratio in the UI; everything before it transfers verbatim. One syntax change to note: Midjourney’s old --cref character-reference flag is V6/Niji only. On V7 and V8.1 use Omni Reference --oref [image-url] with --ow (omni-weight, 0–1000, default 100) to keep the same character across shots.

TL;DR: Never stop at ancient Chinese. Lock five layers in every prompt: dynasty, material, ornament, environment, art style. Use English dynasty names (not pinyin), negative-prompt kimono, japanese to stop drift, and skip dragon, phoenix, lotus unless you want them mangled. Templates work on Midjourney V8.1, Nano Banana 2 and GPT Image 2 alike; drop the --ar flag on the last two.

What a high-quality prompt should contain

Stable Chinese-aesthetic portraits need these five layers. “Ancient Chinese” alone is not enough:

  • Dynasty: Tang Dynasty / Song / Ming / Qing. Clothing, color and hair differ wildly between them.
  • Material: silk / linen / lacquered armor / embroidered satin. Controls texture.
  • Ornament: hairpin and pearl ornaments / gold embroidery / jade earrings.
  • Environment: ink-wash mountains / wooden courtyard / paper lantern. A specific scene, not “a palace”.
  • Art style: painterly cinematic / traditional ink-wash / modern Chinese animation.

10 copy-ready prompt templates

1. Tang Dynasty court lady

Best for: Hanfu portraits, period novel covers

A young Tang Dynasty court lady in flowing red and gold silk hanfu, elaborate hairpin and pearl ornaments, soft round face with rouge, holding a paper fan, warm interior light, ink-wash mountains painted on the wall behind, cinematic --ar 2:3

2. Song Dynasty scholar

Best for: Tea / culture brands, literary themes

A Song Dynasty scholar in restrained indigo robes, quiet thoughtful expression, holding a brush over rice paper, dim candle light, ink-wash mist in background, refined elegant aesthetic, 50mm f/1.4 cinematic --ar 4:5

3. Ming Dynasty general

Best for: Period drama, tabletop cards

A Ming Dynasty general in lacquered black-and-gold armor, stern face, holding a long spear, dramatic side light, war banners blowing in the background, painterly Chinese cinematic style, --ar 2:3

4. Xianxia heroine

Best for: Film promo, game splash

A young xianxia swordswoman on a stone bridge, white silk robes flowing in wind, blue-grey misty mountains, soft volumetric light, modern Chinese cinematic style, slight motion blur on cloth, --ar 16:9

5. Ink-wash landscape figure

Best for: Tea packaging, Eastern art posters

A lone figure on a mountain ridge, traditional Chinese ink-wash painting style, sparse brushstrokes, mist between peaks, calligraphy seal in corner, monochrome black and grey, slight rice-paper texture, --ar 3:4

6. Modern hanfu lifestyle

Best for: Guochao brands, Xiaohongshu posts

A modern young Chinese woman in updated hanfu with sneakers, sitting in a sunlit traditional courtyard, gentle smile, modern lifestyle photography blended with classical setting, 50mm f/1.8 --ar 4:5

7. Qing noblewoman

Best for: Palace dramas, premium brand visuals

A Qing Dynasty noblewoman in vibrant embroidered changshan, jade earrings, sitting in a wood-carved chair, soft northern window light, refined editorial style, 85mm portrait, --ar 4:5

8. Daoist sage

Best for: Cultivation themes, incense / tea brands

A Daoist priest with white beard on a mountain peak above a sea of clouds, flowing grey robes, bamboo staff, sunrise warm light, painterly Chinese style with ink-wash elements, --ar 16:9

9. Chinese wedding

Best for: Bridal brands, wedding ads

A young woman in Han-style ceremonial wedding red, intricate gold embroidery, traditional phoenix crown, soft candle light, painterly editorial wedding photograph, --ar 4:5

10. Tang Dynasty poet

Best for: Literary themes, period games

A Tang Dynasty male poet drinking wine under a willow tree, half-tipsy smile, scroll of poetry on the table, paper lantern overhead, warm ink-wash background, painterly historical scene, --ar 4:5

Common mistakes

  • Writing only ancient Chinese woman, hanfu: dynasty drifts and garments mix.
  • Combining xianxia with palace-drama in one prompt: two different genres, the model splits the difference badly.
  • Spamming dragon, phoenix, lotus ornaments: the model warps repeating motifs. Negative-prompt them and add detail in post.
  • No color grade specified: defaults to gacha-game gold spam. State the grade (muted indigo and ash, warm tungsten).
  • Using pinyin (gufeng, xianxia alone): English dynasty names are far more reliable anchors across all three 2026 models.

How to push results further

  • For ink-wash mood: ink-wash background, sparse brushstrokes, rice-paper texture
  • For Chinese-cinematic look: modern Chinese cinematic, painterly, soft volumetric light
  • Hanfu specifics: flowing sleeves motion, layered silk beats beautiful hanfu
  • Night scene atmosphere: paper lantern, warm tungsten glow
  • For guochao (modern × traditional): modern young Chinese, updated hanfu, sneakers, sunlit courtyard

FAQ

Q: Why does my Chinese portrait look Japanese?

A: Add hanfu and negative kimono, japanese. Models confuse East-Asian costume; you need explicit separation.

Q: Want period-cinematic, not classical painting — how?

A: Skip traditional Chinese painting. Write modern Chinese cinematic, painterly, soft volumetric light, cinematic key visual.

Q: Embroidery patterns come out scrambled — fix?

A: AI struggles with complex repeating patterns. Describe color blocks (solid red silk with gold trim at hem) and layer ornament in post.

Q: Model keeps adding dragons / phoenixes — how to stop?

A: Negative-prompt dragon, phoenix, mythical creatures. If you want subtle motifs, specify subtle cloud pattern only.

Q: How do I keep the same face across a series of shots?

A: On Midjourney V8.1, generate one portrait you like, then reuse it with Omni Reference: --oref [image-url] --ow 100 (raise --ow toward 1000 for stricter likeness). The older --cref flag only works on V6/Niji. On Nano Banana 2 and GPT Image 2, attach the reference image in the chat and ask to “keep the same character, change the scene.”

Q: Which model is best for hanfu portraits in 2026?

A: For a painterly key-visual look, Midjourney V8.1 (native 2K). For sharpest skin and silk micro-texture, Google Nano Banana 2 (true 4K, free users get ~20 images/24h). For clean embroidery patterns or readable Chinese text on banners, GPT Image 2, whose “Thinking” mode reasons about layout before drawing. All take the same prose prompts below.

Tags: #Chinese-style #Image generation #Prompt #Portrait