產品換背景
給 GPT Image 2.5 Sunburst 一張產品照片,說明想換成什麼房間、桌面或光線。把輪廓、標籤與比例等保留要求寫清楚,讓修改圍繞同一件產品進行,而不是重新設計產品。

改變場景,同時保留主體。當需要保留的細節與想修改的地方同樣重要時,GPT Image 2.5 Sunburst 可以協助你生成與調整圖片。
GPT Image 2.5 Sunburst 支援文字與圖片輸入,可以根據描述生成圖片,也能針對參考圖修改。提供產品照片,再說明新的場景、光線,以及要保留的形狀與顏色,想改什麼會更清楚。
模型支援 1024 × 1024、1536 × 1024、1024 × 1536 等像素尺寸。無論是方形產品圖、橫向封面或直向海報,都可以先規劃主體和文字所需的空間。
Sunburst 支援 low、medium、high、extra high、max 五種畫質級別,以及自動選擇。模型著重精細生成與編輯,適合需要仔細處理材質、光線與局部細節的產品圖和創意畫面。
給 GPT Image 2.5 Sunburst 一張產品照片,說明想換成什麼房間、桌面或光線。把輪廓、標籤與比例等保留要求寫清楚,讓修改圍繞同一件產品進行,而不是重新設計產品。

先決定標題放在哪裡,再安排畫面。提供準確文案、資料順序與位置,讓 Sunburst 一起探索文字和視覺的搭配;也可以留下乾淨的空白區域,方便後續排版。

一張參考圖確定主體,一張提供配色,另一張說明場景。告訴 Sunburst 每張圖負責什麼,就能把零散的視覺參考整理成清楚的創作方向。

說明主體、場景、構圖及預期用途。若進行編輯,請指出應變更之處,並列出必須保留的細節。
上傳最多五張 JPEG、PNG 或 WebP 格式的參考圖片,每張大小不超過 20 MiB。解釋每張參考圖片的作用,避免讓模型自行猜測。
選擇「在 YouMind 中試用」,將創作說明和參考圖片帶入已選定 GPT Image 2.5 Sunburst 的對話,繼續生成圖片。長寬比和畫質設定屬於對話中的圖片工作流程,此模型頁面的輸入欄不提供這些設定。
檢查主體、佈局及任何文字。要求進行重點修正,並在進行下一次更改前與先前版本進行比較。
先看手上的任務:精細修改產品圖、探索日常配圖,或是根據已有圖片發展新的視覺方向。
橫向捲動查看所有模型
瀏覽 GPT Image 2.5 Sunburst 提示詞,為你的下一張圖片尋找創作起點。

Using the provided reference image as the character base, generate a 3x3 grid of 9 distinct square panels. Each panel should depict the same character in different candid, slice-of-life scenarios, styled to look like casual snapshots or social media posts.
Layout: A perfectly aligned 3x3 grid with thin white borders separating the panels.
Style & Atmosphere: The overall aesthetic should be soft, slightly grainy, and naturalistic, mimicking amateur photography. Use varied lighting conditions (daylight, indoor warm light, dusk) and camera angles (close-ups, medium shots, full-body shots).
Panel Content Instructions:
1. Top Left: Motion blur effect, character looking back over shoulder outdoors.
2. Top Center: Character sitting at a cafe table with an iced coffee drink in the foreground, resting chin on hand.
3. Top Right: Close-up selfie angle, partially obscured by a finger or thumb covering part of the lens.
4. Middle Left: Character making a peace sign near their face, winking or squinting against bright sunlight.
5. Middle Center: Silhouette shot against a bright window, standing indoors.
6. Middle Right: Low-angle street shot, character walking past a concrete wall.
7. Bottom Left: Extreme close-up of face, mouth open in surprise or mid-speech.
8. Bottom Center: Night/dusk scene, character walking away down a dimly lit street with streetlights.
9. Bottom Right: Rear view of character walking along a sidewalk next to a fence.
Constraints: Maintain consistent character design (hair color, outfit details like the oversized dark hoodie/sweater and shorts) across all panels. Ensure high variety in composition and mood.

A pixel art sprite of a curved banana, rendered on a visible white grid background with black outlines. The image is constructed from square pixels using a limited color palette: dark brown for the stem and shadowed edges, bright yellow for the main body, and orange for highlights and shading. The banana curves diagonally from the top right corner down to the bottom left.

Create a vertical collage of 10 distinct Mid-Autumn Festival posters arranged in a 2-column grid. The style is elegant, poetic, and minimalist Chinese aesthetics.
Top Left: Dark background with yellow osmanthus flowers and a glowing full moon. Text: "花好月圆" (Blooming Flowers & Full Moon), "人间至味是团圆" (The best taste in life is reunion).
Top Right: Ink wash painting style mountains and water with a small boat. Orange sun/moon. Text: "但愿人长久,千里共婵娟" (May we all be blessed with longevity...).
Row 2 Left: Deep blue background showing phases of the moon in a horizontal line. Text: "月亮走过的路也是回家的路" (The road the moon traveled is also the road home).
Row 2 Right: Close-up photography of a mooncake on a ceramic plate with leaf shadows. Text: "时光有味,月饼知秋" (Time has flavor, mooncakes know autumn).
Row 3 Left: Beige background with abstract gold and white circles representing the moon over water ripples. Text: "一轮明月,照见心安" (A bright moon illuminates peace of mind).
Row 3 Right: Red wall with traditional lattice window shadows. Text: "中秋" (Mid-Autumn), "月下有光,家中有暖" (Light under the moon, warmth at home).
Row 4 Left: Green background featuring the back of a white rabbit's head looking up. Text: "把思念轻轻放在月亮上" (Gently place thoughts on the moon).
Row 4 Right: Soft purple/lavender background with hanging paper lanterns. Text: "灯火映圆月,万家皆可期" (Lantern lights reflect the full moon...).
Bottom Left: Night landscape with city lights below and a large orange moon rising over mountains. Text: "山河远阔,人间烟火,皆是团圆" (Vast mountains and rivers... are all reunion).
Bottom Right: Dark moody shot of incense smoke curling upwards from a bowl. Text: "中秋不止一日,而是长久的惦念" (Mid-Autumn is not just one day...).

Referencing the uploaded photo of the old house, precisely replicate it as a museum-grade ultra-realistic physical miniature model at 1:24 scale, strictly restoring architectural structure, roof, doors/windows, walls, and original environment. Finely present traces of time such as mottled wall peeling, weathered bricks/tiles, cracks, moss, dust, etc., with realistic natural materials and extreme handmade texture. The model is placed on an exquisite rectangular sand table base, viewed from a 45-degree overhead three-quarter angle, fully displaying the building's appearance. Professional macro photography, soft natural light, realistic shadows, neutral gray studio background, 8K ultra-high definition. Emphasize realistic physical model effect, avoiding plastic toy feel, cartoon feel, and 3D rendering feel.

[{argument name="dynasty" default="Ancient"}] Women's undergarments × Layering steps × Structural unfolding × Materials and cuts × Costume restoration × Vertical 9:16

將上載的相片轉化為可愛的極簡手繪塗鴉插畫。保留原本的人物、姿勢、構圖及可辨識的面部特徵,但將其簡化成迷人的卡通造型。
採用樸拙的手製速寫本風格,以粗獷而略帶搖晃感的鈷藍色毛筆勾線、簡單的圓潤形狀、極少的細節,以及不完美的手繪筆觸呈現。在臉頰加上淡淡的暖桃色/粉紅色紅暈,並在人物周圍點綴細小的金黃色心形圖案。
保持插畫乾淨通透,配以溫暖的米白色/奶白色紙張背景。使用扁平且極為有限的色彩,不加寫實陰影、漸變或複雜紋理。整體感覺應甜美、純真、活潑、溫馨又浪漫,猶如小朋友親手繪製的塗鴉或私人筆記本上的隨筆。
風格關鍵詞:極簡塗鴉、民俗風手繪、藍墨水插畫、可愛情侶插畫、樸素藝術(naive art)、不完美筆觸、簡單卡通、手製速寫本美學、暖米色紙張、淡淡紅暈、細小心形塗鴉、迷人的不完美、乾淨構圖。

Insert {argument name="surreal_subject" default="T-Rex"} into a Song Huizong flower-and-bird painting.
General Formula:
Song Huizong aesthetics × Song Academy Gongbi style × [{argument name="surreal_subject" default="T-Rex"}] × Extreme Gongbi details × Slender Gold script poem × Vermilion seal × Song-style negative space × Ancient silk texture × Deep blue square seal in bottom-left corner with 'Deep Blue' on top and 'DeepBlue' below

Make it an ink wash painting with a brushstroke style that has the beauty and transience of calligraphy, abstracted like cursive script.
Ignore details; aim for a beauty where the movements are understandable even if abstract.

Using REFERENCE_1 as the character base, create a {argument name="aspect ratio" default="16:9"} image made of exactly 9 equal panels in a 3×3 grid, like energetic Vocaloid music-video cuts. Keep the character design consistent, but vary pose, camera angle, crop, facial expression, and background in every panel. Use a flat, clean anime MV style with bold red, blue, cyan, and white graphic shapes, crisp line art, high contrast, and no UI, captions, watermark, or screenshot frame.
Panel plan, exactly 9 cuts:
1. Extreme close-up crop of the character’s eyes and bangs with a blue-and-white spiral graphic background.
2. Full-body dynamic jump/dance pose on a red-and-white checkerboard floor against a deep blue background.
3. Side-profile close-up with eyes closed, hair flowing dramatically, and a large white circle on a red background.
4. Chibi-like forced-perspective shot reaching one gloved hand toward the camera inside a bold blue-and-white circular graphic.
5. Front upper-body idol pose, winking and pointing both fingers toward the cheeks, with red, cyan, and white burst shapes and sparkles.
6. Three-quarter back view looking over the shoulder in front of a blue nighttime city window scene.
7. Close-up of the red sneakers and legs running or stepping across cyan speed lines.
8. Tiny full-body running figure across floating white piano-key platforms, with red circles and cyan audio-wave graphics on a dark blue background.
9. Happy close-up reaching one gloved hand toward the viewer, smiling with eyes closed, surrounded by red and blue confetti.
Overall direction: make the grid feel like a polished animated opening or Vocaloid MV storyboard for {argument name="character name" default="this character"}; preserve the reference character’s identity while prioritizing varied cinematic cuts and rhythmic graphic backgrounds.

Turn this structure photo into a single high-end halftone architectural travel-collage poster. Output one poster per uploaded photo — never merge multiple structures or photos into one frame.
FORMAT
3:4 vertical canvas. Split into two roughly equal zones — real architectural photo on top, printed reconstruction below — divided by one clean, sharp horizontal line.
Do not stretch, warp, tilt, or unnaturally reposition the structure to force the proportions.
TOP HALF — THE REAL SHOT
Preserve the original structure faithfully: same architecture, proportions, perspective, distinctive details, surrounding elements, natural light and shadow, and original color mood.
Keep recognizable features such as:
* façade
* windows and arches
* towers or domes
* doors and structural details
* roofline
* surrounding landscape or street elements
Add only a subtle high-end color grade and the faintest film grain. Do not redesign or reinterpret the architecture.
BOTTOM HALF — THE PRINTED RECONSTRUCTION
Rebuild the same structure as an experimental printed architectural collage using:
torn photographic fragments + fixed-width ASCII character fields + photocopy halftone dots + architectural linework.
The structure must remain immediately recognizable.
* Torn-edge photographic fragments preserve the most important architectural details.
* Brick, stone, glass, metal, shadows, windows, and surrounding textures gradually dissolve outward into ASCII characters, black halftone dots, fragmented print grain, and fine architectural drafting lines.
* Use thin technical construction lines to subtly trace the structure’s geometry.
* Let selected areas break apart into paper fibers and halftone fragments, creating the feeling that the building is emerging from an archival print.
* Keep the main silhouette strong and readable.
* Texture: rough white paper-fiber edges, subtle photocopy imperfections, slight print misregistration, faded ink, old-paper stains, a few restrained crop marks, and one small red calibration bar.
* Color: warm gray old-paper background + black ink + the structure’s original muted colors + a single restrained red accent.
* Never fully desaturate the structure.
The structure should occupy roughly 60–88% of the frame width, scaled according to what reads best for the architecture.
Leave 22–38% of the paper visibly untouched and empty, creating generous negative space so the poster feels like an archival architectural specimen rather than a packed digital screen.
TYPOGRAPHY
Add one clean, readable fixed-width architectural title based on the structure:
[STRUCTURE NAME]
Below it include:
[CITY, COUNTRY]
REC. STUDY 01
Add one short observational caption related to the architecture, such as:
“A monumental study in geometry, time, and place.”
Typography should feel like an old architectural archive mixed with an underground experimental design journal.
A few loose ASCII glyphs may trace the outer silhouette or architectura

A reference sheet for a Japanese text style, divided into six panels. The style is bold, glossy, and colorful with thick outlines. Panel 1 (top left) shows the word "サムネ" in white text with a purple outline and drop shadow. Panel 2 (top center) shows the character "の" in white with a pink circular background. Panel 3 (top right) displays "フォント" in pink with a darker pink outline. Panel 4 (bottom left) features "透過設定" in white with a red outline. Panel 5 (bottom center) shows "練習" in pink with a dark pink outline. Panel 6 (bottom right) combines all elements: "サムネ", "の", "フォント", "透過設定", and "練習" arranged together.

{argument name="theme" default="Dream of the Red Chamber: Night Banquet of Flowers"} × {argument name="reference_composition" default="The Last Supper classic 13-person group portrait composition"} × Borrowing only the horizontal group structure × Approx 1.9:1 aspect ratio × Left side 6 people: {argument name="left_characters_list" default="Miaoyu, Wang Xifeng, Shi Xiangyun, Tanchun, Li Wan, Lin Daiyu"} × Center: {argument name="center_character" default="Jia Baoyu"} × Right side 6 people: {argument name="right_characters_list" default="Xue Baochai, Yingchun, Xichun, Ping'er, Xiren, Qingwen"} × Horizontal long table banquet × {argument name="art_style" default="Qing Dynasty court Gongbi figure painting"}
Instructions:
The "Last Supper 13-person classic group portrait composition" borrows ONLY its central figure + 6 figures on each side + horizontal long table + left/right grouping relationship. The image strictly maintains 6 people on the left, 1 in the center, and 6 on the right, totaling 13 people. {argument name="center_character" default="Jia Baoyu"} is located in the exact center of the frame, with {argument name="left_adjacent_character" default="Lin Daiyu"} and {argument name="right_adjacent_character" default="Xue Baochai"} immediately adjacent to {argument name="center_character" default="Jia Baoyu"} on either side respectively; other characters expand towards the sides according to the order specified in the prompt. All 13 people are clearly visible; do not add or remove characters, do not shuffle the left-right order, and do not disperse characters to other areas. Characters form a complete horizontal group portrait along both sides of the long table. Borrow ONLY the compositional structure of "The Last Supper"; the story, characters, costumes, scenes, and painting language all adopt the classical Chinese context of "Dream of the Red Chamber", with no Western religious figures or religious elements appearing.