OpenZhenZhen AI 提示词画廊 · gpt-image-2.5 · image
忠实自然照片修复与细节增强
Faithful Natural Photo Restoration and Detail Enhancement
Subtle faithful restoration that retains identity, composition and photographic character without unsupported output or tool-parameter promises. / 克制忠实修复，保留身份、构图和摄影特征，不承诺未经支持的输出规格或工具参数。
提示词 / Prompt
中文提示词
将已配置的 identity_reference ref-10a-v1 作为完整待修复图像。忠实修复并增强这张提供的照片。严格保留原身份、面部结构、比例、取景、透视、构图、物体和摄影特征；不要重新诠释场景或重塑面部。恢复自然准确的细节，同时保留真实皮肤纹理。执行subtle, faithful cleanup with natural detail recovery、retain natural skin and surface texture; reduce only clearly accidental compression blemishes、balanced exposure and neutral natural color without dramatic recoloring和reduce only obvious compression artifacts or sensor noise when separable from real detail。目标为高保真、超写实的摄影清晰度、锐利但自然的焦点、专业摄影质感、带细微体积光的现代电影照明、平衡调色、自然景深和 HDR 式明暗范围。Arri Alexa 原始摄影质感仅作视觉参考，不表示图片实际由该设备拍摄或将由其拍摄。在现有图像与工具能力范围内尽量提升感知细节；不要承诺 4K 输出、特定放大器、采样器、步数、CFG 数值或降噪强度。避免创意重绘、风格改变、身份更改、面部重塑、特征夸张、卡通或绘画效果、过度锐化、塑料感皮肤、模糊、新增噪点或胶片颗粒、压缩伪影、失真、解剖错误、过曝、欠曝和褪色。不要增删物体、裁切、添加文字或水印，也不要虚构细节。
English prompt
Use the configured identity_reference ref-10a-v1 as the complete image to restore. Faithfully restore and enhance this provided photograph. Preserve the original identity, facial structure, proportions, framing, perspective, composition, objects and photographic character exactly; do not reinterpret the scene or reshape the face. Recover natural, accurate detail while retaining realistic skin texture. Apply subtle, faithful cleanup with natural detail recovery, retain natural skin and surface texture; reduce only clearly accidental compression blemishes, balanced exposure and neutral natural color without dramatic recoloring and reduce only obvious compression artifacts or sensor noise when separable from real detail. Aim for high-fidelity, ultra-realistic photographic clarity, sharp but natural focus, professional photographic finish, modern cinematic lighting with subtle volumetric light, balanced color grading, natural depth of field and HDR-like tonal range. An Arri Alexa raw-photo aesthetic is a visual reference only, not a claim about the camera that captured or will capture the image. Maximize perceptual detail within the available image and tool; do not promise 4K output, a particular upscaler, sampler, step count, CFG value or denoising strength. Avoid creative reinterpretation, style changes, identity alteration, face reshaping, exaggerated features, cartoon or painterly rendering, over-sharpening, plastic skin, blur, added noise or film grain, compression artifacts, distortion, anatomical errors, overexposure, underexposure and washed-out color. Do not add or remove objects, crop, add text or watermark, or invent details.
探索更多图片和视频提示词 · 生成图片 · 生成视频 · 在 AI 画布中创作
相关作品
- 忧郁气质哑光搪胶手办微缩场景
Transform the one referenced person into a single softly melancholic designer-vinyl figure, preserving visible identity and outfit cues, in a refined moonlit miniature product scene. The simple dark lower outfit is an explicit toy-design default because it is outside the reference crop. / 将唯一参考人物转化为
- 沧海孤崖望无极
东方水墨山海武侠风，16:9横版史诗构图。前景一位孤独侠客背对观者，站在临海悬崖尽头，腰佩长剑，衣袍被海风吹起，静静望向远方。画面中央保持开阔，视线能够直接穿过广阔海面，一直延伸到极远处清晰的海天交界线，形成强烈的“望向无限远方”的空间纵深感。海面辽阔苍茫，波涛起伏，分布着若干古代帆船和楼船，船只远近递减，强化空间尺度。 远景一侧矗立着一棵真正震撼天地的通天巨树。通天树位于画面左侧或右侧远方，扎根于遥远海中的巨大岛屿和山体之上，与前景侠客之间隔着广阔海域。重点强调：通天树高到超出画面，不能看到树顶，树冠的主体部分在画面外，只能看到靠近下沿的一部分最底层树枝和树冠边缘。整棵树继续向天空无限延伸
- 贴地低视角羽毛球动作人像
A photorealistic vertical sports portrait of the referenced woman on an indoor court, captured in a low kneeling lunge with one racket and shuttlecocks in motion. / 室内球场上的写实纵向运动人像：参考人物以贴地低视角单膝跪姿挥拍，羽毛球掠过前景。
- 幻想矮人特效化妆十二步过程图
A consistent 3-by-4 makeup-studio storyboard showing the same referenced person through twelve cumulative practical-effects steps into an original humanoid fantasy dwarf. / 在化妆工作室以3列×4行分镜展示同一参考人物逐步完成十二个累积特效化妆步骤，变身为原创类人幻想矮人。
- 早餐店飞食定格商业动作人像
A photorealistic breakfast-shop action portrait using the actual configured identity, with all source-requested wardrobe, motion and negative constraints. / 使用实际配置身份的写实早餐店动作肖像，保留来源指定服装、飞食动作与负向约束。
- 电影感无边框五视角街头人像拼贴
Five expressive views of the same identity in a coherent cinematic borderless collage. / 同一身份的五种表现视角组成连贯电影感无边框拼贴。