Galerie de prompts IA OpenZhenZhen · gpt-image-2.5 · image

Portrait bleu-vert et poésie de papier découpé : moitié inférieure

Blue-Green Portrait and Papercut Poetry: Lower Half

Portrait bleu-vert et poésie de papier découpé : moitié inférieure — découvrez cette création avec gpt-image-2.5 sur OpenZhenZhen, son prompt original et un exemple de génération à réutiliser.

Portrait bleu-vert et poésie de papier découpé : moitié inférieure — AI image creation
Portrait bleu-vert et poésie de papier découpé : moitié inférieure

Prompt

Prompt chinois

将完整上传参考照片同时作为原始图像内容和其中成年女性的面部身份依据。保持整张原照片完整、不裁切且清晰可辨，并据此创作一张自适应社论艺术作品。 来源只规定下半部分的处理，没有指定上半部分处理方式或最终画幅比例。本次实现明确采用以下版式默认：将完整的 3027×1531 原照片不作改动地放在上半部分，按原始未裁切比例填满横向画格；下半部分设置为等高画格。由此形成约 3027×3062 像素的版式描述，比例约 0.989:1；这些尺寸只是设计说明，不是渲染器参数。不得裁切、放大局部、重绘、改色或替换原图任何部分。准确保留照片中女性的面部、深色长发、纤细黑色肩带和青绿色背景。不要推断地点，也不要添加道具。 下半部分只转译照片中确实可见的视觉特征。以深色发丝为灵感画出几道流动弧线，用一条纤细黑线呼应可见肩带，并从背景提取克制的青绿色色块；将它们化为少量手工剪纸形状、柔和的不规则轮廓和手绘标记。元素保持简洁、有触感、松散排列，并留出充足的白色或暖纸色空间；本次构图默认让下半画格约 70% 保持留白。不要添加人物、物体、景色、建筑、天气、产品或任何与照片无关的符号；下半部分也不要照字面重画原场景。 只加入一条小巧且源自图像的诗意短句“静谧，留在青绿之间”，以轻柔手写或排版方式置于下半部分。文字安静融入画面，像私人视觉日记注释，而非常规海报信息。采用微妙纸张颗粒、轻印刷或盖印质感、松散几何和有意的不对称。下半部分应有趣、诗意、略带稚拙、自然即兴，同时概念清晰。避免僵硬网格、刻意对称、杂乱、光亮矢量图形、将下半部分照相写实地重绘、不可读文字、标志或水印。 上传的完整照片就是实际 ref-10a-v1 身份文件；同一份完整参考同时作为原始图像输入，不另加第二张来源照片，也不裁切。

Prompt anglais

Use the complete uploaded reference photograph as both the original image source and the facial identity reference for its visible adult woman. Keep the original full photograph uncropped and recognizable. Create one adaptive editorial artwork based on this exact image. The source specifies treatment for the lower half but does not define an upper-half treatment or final canvas ratio. For this implementation, use this clearly stated layout default: place the complete original 3027×1531 photograph unchanged in the upper half, filling a horizontal panel at its native uncropped proportions. Add a lower panel of equal height. The resulting canvas is approximately 3027×3062 pixels in layout terms, about 0.989:1; treat these measurements as a design description, not a renderer parameter. Do not crop, zoom into, redraw, recolor or replace any part of the source photo. Preserve the woman’s actual face, long dark hair, thin black shoulder straps and the teal background exactly as visible. Do not infer a location or add props. For the lower half, interpret only visual qualities that are genuinely present in the photo. Translate a few flowing arcs inspired by her dark hair, one fine black line echoing the visible shoulder strap, and a restrained teal field sampled from the background into a small group of handmade cut-paper shapes, softly imperfect silhouettes and hand-drawn marks. Keep the elements simple, tactile and loosely arranged, with generous white or warm-paper negative space; as a composition default, leave roughly 70% of the lower panel open. Do not add a person, objects, scenery, architecture, weather, products or symbols unrelated to the visible image. Do not literally reconstruct the photographed scene in the lower panel. Add only one small, context-derived poetic note, “Quiet, held in teal”, lightly handwritten or typeset in the lower panel. Keep the typography quiet and integrated, like a private visual-diary annotation, not conventional poster information. Use subtle paper grain, light print or stamp texture, loose geometry and thoughtful asymmetry. Make the lower half feel playful, poetic, slightly naïve, spontaneous and conceptually intentional. Avoid rigid grids, calculated symmetry, clutter, glossy vector graphics, photorealistic redraw of the lower panel, unreadable text, logos and watermarks. The uploaded full photograph is the same actual ref-10a-v1 file used for facial identity. It remains one complete reference input; no separate source photo or crop is introduced.

Voir le modèle API

Découvrir plus de prompts image et vidéo · Générer des images · Générer des vidéos · Créer sur le canevas IA