Skip to content
Pochime's LINE Sticker Site

Keqing AI Image Prompts: 5 Distinct Poses and Results

15 min read
Keqing AI Image Prompts: 5 Distinct Poses and Results — Generate Keqing in five deliberately different poses, settings, and camera systems. Includes exact English prompts, Japanese translations, a shared negative prompt, and result analysis.

Generate Keqing in five deliberately different poses, settings, and camera systems. Includes exact English prompts, Japanese translations, a shared negative prompt, and result analysis.

This experiment designs five non-overlapping scenes for a social-ready Keqing AI illustration series.

What this article covers

  • 1A reusable identity prompt for Keqing
  • 2Five non-overlapping poses, settings, and camera systems
  • 3English prompts with Japanese translations
  • 4A shared negative prompt
  • 5Actual outputs and social-thumbnail analysis

1What stayed fixed and what changed for Keqing

Lavender cat-ear-styled twin tails, violet eyes, and the ornate purple-white gold-trimmed outfit remain fixed. A crane-beam vault, harbor-model planning session, rice-terrace survey, monumental bridge inspection, and lantern-shadow route change velocity, camera, scale, environment, and the meaning of her Electro motif.

  • Keep recognizable hair, eyes, outfit, and signature ornaments fixed
  • Give every concept a different body axis and large silhouette
  • Change setting, time, and key-light direction on every run
  • Separate low angle, aerial view, profile, symmetry, and lateral tracking
  • Keep the face and action readable at social-media thumbnail size

Shared negative prompt for all five concepts

This shared avoid list targets duplicate subjects, anatomy failures, bad crops, unwanted text, and a return to repetitive compositions.

text, logo, watermark, signature, duplicate character, duplicate reflection, extra limbs, duplicate limbs, extra fingers, fused fingers, malformed hands, distorted anatomy, cropped face, hidden violet eyes, dark hair, missing twin tails, muddy colors, low detail, photorealism, 3D render, sexualized pose

3Five Keqing image prompts and results

The prompt panel opens in English and can be switched to Japanese. On desktop it scrolls inside a rounded panel matched to the adjacent image height; on mobile it expands so normal page scrolling continues.

Result 1: harbor crane-beam vault

Harbor foreshortening
AI-generated Keqing vaulting along a cargo crane beam above Liyue Harbor

Cargo-crane action prompt

Premium vertical anime action key visual of Keqing from Genshin Impact, faithful recognizable official design: young fair-skinned woman, long lavender twin tails styled like cat ears, violet eyes, ornate purple-and-white short dress with gold details, dark gloves, slim sword. At blue dawn in Liyue Harbor she sprints along the top of a moving cargo crane beam, plants one boot, and vaults sideways over a swinging empty hook while drawing her sword in one controlled motion; rainwater and violet sparks trail behind without a large lightning blast. Extreme low-angle side-tracking wide lens from the dock, foreshortened boot, crane beam and sword edge in foreground, teal harbor, red sails and warm warehouse lights below, decisive expression, crisp premium social-media anime illustration. One adult character only; hook is empty, no other people or creatures, no text, logo, watermark, signature, duplicate character, extra limbs, malformed hands, fused sword and body, cropped face, photorealism, 3D render, or sexualized pose.

Result 2: harbor model planning

Civic design close-up
AI-generated overhead Keqing planning bridges with a harbor model and abacus

Yujing Terrace planning prompt

Premium vertical anime editorial illustration of Keqing from Genshin Impact, faithful recognizable official design: young fair-skinned woman, long lavender twin tails styled like cat ears, violet eyes, ornate purple-and-white dress with gold details, dark gloves. Before sunrise in a quiet Yujing Terrace planning room she leans over a circular table, moving one tiny bridge segment into a scale model of Liyue Harbor with one hand while the other slides beads on a small wooden abacus; rolled plans are tied shut with blank ribbons. Strict directly overhead composition, face, hands, bridge model and abacus clearly readable, no sword, combat or electro effects, warm desk lamp against cool window light, efficient civic focus, refined crisp anime art. One adult character only; no other people or creatures, no readable writing or map labels, no text, logo, watermark, signature, duplicate reflection, extra limbs, extra fingers, malformed hands, cropped face, photorealism, 3D render, or sexualized pose.

Result 3: sunrise terrace survey

Practical profile
AI-generated side profile of Keqing surveying an irrigation gate in rice terraces

Rice-terrace survey prompt

Premium vertical cinematic anime portrait of Keqing from Genshin Impact, faithful recognizable official design: young fair-skinned woman, long lavender twin tails styled like cat ears, violet eyes, ornate purple-and-white dress with gold details. At golden sunrise in steep green rice terraces she stands in strict side profile on a narrow stone path, using a brass survey compass and plumb line to measure the angle of one small irrigation gate while her twin tails stream downhill in the breeze. Eye-level 135mm full-body composition, no sword, combat or electro magic, mirrored terrace water, fresh green rice, lavender hair and warm gold rim light, practical attention and quiet optimism, refined premium anime art. One adult character only; no farmers, other people, birds or creatures, no readable markings on tools, no text, logo, watermark, signature, duplicate reflection, extra limbs, malformed hands, cropped body, photorealism, 3D render, or sexualized pose.

Result 4: gorge bridge inspection

Monumental engineering
AI-generated distant Keqing inspecting cables beneath a colossal unfinished bridge

Suspended scaffold prompt

Premium vertical cinematic anime illustration of Keqing from Genshin Impact, faithful recognizable official design: young fair-skinned woman, long lavender twin tails styled like cat ears, violet eyes, ornate purple-and-white dress with gold details. In gray morning rain she stands tiny on a narrow suspended bamboo inspection platform beneath an unfinished stone bridge spanning a colossal Liyue gorge, secured by a visible waist safety line, holding one rolled blank blueprint under an arm while checking a taut support cable with the other hand. Very distant lateral telephoto composition, full-body silhouette beneath gigantic arches and scaffolding, river and mist far below, no sword, combat or electro magic, civic duty and monumental engineering, wet gray stone, bamboo brown and one lavender accent, refined premium anime poster art. One adult character only; no workers, other people, birds or creatures, no readable text, logo, watermark, signature, duplicate character, extra limbs, malformed hands, cropped body, broken safety line, photorealism, 3D render, or sexualized pose.

Result 5: lantern-shadow civic route

Planned lightning symbol
AI-generated bird's-eye view of Keqing within a lightning-shaped lantern route

Empty civic plaza prompt

Premium vertical conceptual anime poster of Keqing from Genshin Impact, faithful recognizable official design: young fair-skinned woman, long lavender twin tails styled like cat ears, violet eyes, ornate purple-and-white dress with gold details. She stands tiny and still at the exact center of an empty Liyue civic plaza before dawn, seen from directly overhead, holding one closed blank planning scroll. Twelve ordinary amber street lanterns and their long shadows align across the stone grid to form one enormous lightning-bolt route around her without any electricity or magical glow. Strict bird's-eye geometry, centered full-body figure, charcoal paving, warm amber pools, violet clothing and pale blue pre-dawn edges, human planning rather than supernatural power, huge graphic negative space, refined anime poster art. One adult character only; exactly twelve lanterns, no electro magic, no other people or creatures, no readable text, logo, watermark, signature, duplicate reflection, extra limbs, malformed hands, photorealism, 3D render, or sexualized pose.

9The priority order that creates visible variety

Changing only a background noun is not enough when the subject remains upright in the center. Separate body axis and action first, then assign camera, time, light, palette, and emotion. This preserves character identity while making the series read as five independent ideas.


10Conclusion: design five differences before generation

Locking only Keqing's identity while separating pose, setting, camera, time, and emotion produced five results that do not feel like near-duplicates. A diversity matrix is more reliable than generating a large random batch and deleting look-alikes afterward.

Frequently asked questions

QHow do I stop AI images of the same character from looking repetitive?

Assign a different body axis, action, camera height, time, and eye path before writing the scenery. Do not center an upright subject in every concept.

QWhy not run the same prompt five times?

Same-prompt batches measure consistency, but they do not guarantee scene diversity. This social series shares only the identity description and deliberately changes the scene design.

QWhich language appears first in the prompt panel?

English is the initial display. Every prompt can be switched to its Japanese translation.

Article tags

Share this article