霓虹故障动漫旅行双联编辑海报

Please turn each character photo I upload into an individual high-end travel portrait editorial poster, no multi-image stitching, output each photo separately. Overall use a 3:4 vertical diptych composition, with the top retaining the clearly visible real original image and the bottom performing a "neon glitch anime portrait" redraw; the upper and lower borders are horizontal, clear, and restrained, with no atomization or mutual penetration.

The upper part accounts for about 45%–50% of the frame, faithfully retaining the original photo. Accurately maintain character identity, quantity, face shape, facial feature relationships, hairstyle, glasses, gaze, expression, posture, gestures, clothing, personal items, and the spatial relationship between characters and the environment. Only perform slight art magazine-style color grading and extremely thin film grain processing, it must still clearly be real photography. To adapt to the frame, the sky, walls, or surrounding environment can be naturally extended, but the character must not be redrawn, faces changed, stretched, mirrored, copied, or the subject moved.

The lower part extracts the most recognizable face, hairstyle, clothing, gestures, and key items of the character from the same photo, reconstructing them into a contemporary anime editorial portrait with strong visual impact. Do not copy the full framing and background density of the original photo; significantly enlarge the character's half-body to occupy about 65%–95% of the lower area's width, which can be moderately cropped out of the frame, letting the line of sight, action direction, and silhouette become dominant. The background retains only one to three architectural lines, lights, railings, furniture, or spatial structures from the original image, compressed into low-information light-color fields and structural fragments.

The redraw uses angular color blocks, clear anime line-plane relationships, deep indigo shadows, steel blue and cyan highlights, and selects one existing warm color from the original image to translate into a restrained coral pink or magenta accent color. Add limited halftones, partial RGB color offset, edge signal breaks, and a small amount of pixel chunks. Glitch effects can only occur at hair edges, shoulder silhouettes, clothing outer edges, and background structures, and must not cover or destroy faces, eyes, hands, and key items; must not change character identity, nor generate irrelevant cyber cities.

The lower area retains about 18%–28% of continuous deep blue or low-information color surface as a breathing area. The text is concentrated into a group, the main title uses bold narrow game title fonts that are clearly readable in normal chat previews, with slight pixel cuts or RGB misalignments; auxiliary fields are smaller, but cannot be shrunk into gray lines.

Extract a 1–3 English word title based on the actually visible location, character state, action, or light in the photo, and add the field "SIGNAL NOTE 01" and a 5–12 English word observation sentence. If the location cannot be confirmed, do not fabricate place names, the title "AFTERLIGHT" can be used, and the observation sentence "A quiet face held inside moving light." Only render confirmed text, do not add author names, brands, accounts, logos, QR codes, or garbled characters.

The overall color is extracted from the original photo and compressed into a limited palette, maintaining the original warm and cold character of the character's clothing and environment. The image should have a sense of real travel atmosphere, anime character tension, video game cover feel, and editorial publication restraint; glitch is rhythm, not obstruction.

Avoid generic cyberpunk, full-screen neon rainbows, dense HUD, UI text too small to read, vaporwave templates, facial glitches, extra limbs, cheap filters, and full-frame illustrationization.

请将我上传的每一张人物照片分别制作成一张独立的高级旅行肖像编辑海报，不多图拼接，每张照片单独输出。整体采用3:4竖版双联构图，上方保留清楚可见的真实原图，下方进行“霓虹故障动漫肖像”转绘；上下边界水平、明确、克制，不雾化、不互相渗透。

上半部分约占画面45%–50%，忠实保留原始照片。准确保持人物身份、数量、脸型、五官关系、发型、眼镜、目光、表情、姿态、手势、衣着、随身物件以及人物与环境的空间关系。只进行轻微的艺术杂志式调色与极薄胶片颗粒处理，仍必须明显是真实摄影。为适配画幅，可自然扩展天空、墙面或周边环境，但不得重画人物、改变面孔、拉伸、镜像、复制或移动主体。

下半部分从同一张照片中提取人物最具识别度的面部、发型、服装、手势和关键物件，重构为具有强烈画面冲击的当代动漫编辑肖像。不要照搬原照片的完整取景和背景密度；将人物半身明显放大，占下区宽度约65%–95%，可适度裁切出画，让视线、动作方向和轮廓成为主导。背景只保留一至三条来自原图的建筑线、灯光、栏杆、家具或空间结构，并压缩成低信息光色场与结构碎片。

转绘使用角面化色块、清楚的动漫线面关系、深靛蓝阴影、钢蓝与青色高光，并从原图中选择一处真实存在的暖色，转译为克制的珊瑚粉或品红跳色。加入有限网点、局部RGB套色偏移、边缘信号断裂和少量像素切块。故障效果只能发生在头发边缘、肩部轮廓、衣物外缘及背景结构，不得覆盖或破坏脸、眼睛、手和关键物件；不得改变人物身份，也不得生成无关的赛博城市。

下区保留约18%–28%连续的深蓝或低信息色面作为呼吸区。文字集中成一组，主标题使用在正常聊天预览中清楚可读的粗窄游戏标题字，带轻微像素切口或RGB错位；辅助字段较小，但不能缩成灰线。

根据照片中真实可见的地点、人物状态、动作或光线提炼1–3个英文单词标题，并加入字段“SIGNAL NOTE 01”与一行5–12个英文单词的观察句。若地点无法确认，不得编造地名，可使用标题“AFTERLIGHT”，观察句使用“A quiet face held inside moving light.”。只渲染已确定的文字，不添加作者名、品牌、账号、Logo、二维码或乱码。

整体色彩从原照片提炼并压缩为有限色盘，保持人物衣物和环境原有的冷暖性格。画面应兼具真实旅途气息、动漫人物张力、电子游戏封面感和编辑出版物的克制感；故障是节奏，不是遮挡。

避免通用赛博朋克、满屏霓虹彩虹、密集HUD、小到不可读的界面字、蒸汽波模板、面部故障、额外肢体、廉价滤镜和全画面插画化。