carat

Prompt Gallery

/

배경

Login
Sign Up

배경

제품 촬영, 인물 사진, 프로젝트를 위한 고품격 배경 이미지를 AI로 자유롭게 생성해 보세요.

2026.07.08

·
GPT 이미지 2로 생성한 그라데이션 배경

그라데이션 배경

GPT 이미지 2· Image
A smooth abstract gradient background transitioning from deep purple to warm orange to soft pink, ethereal flowing color blends, no objects, pure color study., Digital illustration with neon glow effects, neon rim lighting, dark ambient with glowing accents, futuristic, edgy, vibrant mood, electric magenta-cyan duotone with deep black, dynamic dutch angle, atmospheric haze, No text, no watermark.
GPT 이미지 2로 생성한 무지개 배경

무지개 배경

GPT 이미지 2· Image
A vibrant rainbow gradient background with soft pastel rainbow stripes flowing diagonally, dreamy clouds at the edges, gentle and uplifting atmosphere., Modern flat editorial illustration, implied soft directional light, no heavy shadows, contemporary, sophisticated, clean mood, limited 3-4 color palette with one accent, muted tones, rule of thirds, deliberate negative space, No text, no watermark.
GPT 이미지 2로 생성한 줌 화상회의 배경

줌 화상회의 배경

GPT 이미지 2· Image
A clean professional home office background for video calls, bookshelf with plants, warm lighting., Claymation-style 3D render, tactile clay texture, warm softbox studio lighting with visible fingerprints texture, whimsical, handmade, charming mood, saturated but slightly matte colors, fingerprint visible, eye-level close-up, shallow depth of field, No text, no watermark.
gpt-image-2으로 생성한 배경 인물 제거

배경 인물 제거

GPT 이미지 2· Image
Remove all unwanted people from the background while perfectly preserving the main subject, clothing, facial features, pose, and lighting. Reconstruct the background naturally where the removed people were located. Match realistic perspective, lighting direction, shadows, textures, architecture, and depth. Do not crop or stylize the image. The result should look like the original photo was taken without people in the background.
미드저니로 생성한 AI 이미지 예시

핑크 보브 치파오 인물사진

미드저니· Image
Vertical 9:16 photorealistic full-body portrait, nostalgic Taiwanese idol drama style, early-2000s analog flash photography, soft highlight bloom, gentle halation, dreamy blur, film grain, slight noise. A young East Asian woman with a short pink bob haircut and blunt bangs wears a fitted red qipao with subtle floral jacquard texture and matching red high heels. She has a cool, cocky, rebellious expression and looks directly at the camera. Full-body composition, head-to-toe framing, her entire body visible including both legs, feet, and red high heels. Slight low-angle wide-lens perspective to make the legs look long and stylish while keeping realistic anatomy. She stands at the entrance of a nostalgic Taiwanese snack shop with vintage Chinese signage, warm ambient light, old storefront textures, retro street atmosphere, and worn 2000s idol-drama mood. Pose: both arms extended toward the camera, forming the classic C-BLOCK “以下范上” gesture. Left hand higher, palm inward, index and middle fingers pressed together pointing upward, thumb extended sideways. Right hand lower, palm inward, index and middle fingers pressed together pointing downward, thumb extended sideways crossing near the left thumb. Hands are closer to the lens but not too huge, face clearly visible between or above them. Keep fingers straight, natural, and readable. Preserve hand skin texture and clear finger details. Do not overexpose the hands. Avoid loose peace signs, single-finger pointing, claw-like hands, distorted anatomy, cropped feet, hidden shoes, or hands covering the face.
1a3ec7fbfee4702b8f550ef8d9c98b7d
미드저니로 생성한 AI 이미지 예시

모스번 픽셀 펫 캐릭터

미드저니· Image
Create a single original cute fantasy creature for a cozy collectible card game. Creature number: 1. Creature name: Mossbun. Style: pixel art sprite, retro handheld game aesthetic, crisp pixel edges, no anti-aliasing, no gradients, no soft shading, no smooth painterly rendering. Use exactly two visible colors only: #ffffff and #8a8a8a. Transparent background is allowed. No black, no colored pixels, no additional gray shades, no highlights in other colors. Full body, centered, small magical pet-like creature, cute and collectible, clean readable silhouette, charming expression. The creature should be unique and recognizable with a few distinctive fantasy details, but simple enough to read clearly at small size. No environment, no scenery, no frame, no card border, no text, no logo, no watermark, no UI elements. Output: square pixel art sprite, transparent background, designed for 128x128 pixels.
9936745fd486c037d9304599cfe37a9c
GPT 이미지 2로 생성한 밤하늘 배경

밤하늘 배경

GPT 이미지 2· Image
A breathtaking night sky filled with thousands of stars, the Milky Way galaxy visible across the center, deep blue and purple hues, silhouetted mountains at the bottom., High-detail 3D render, Octane-style, dramatic volumetric lighting with rim light, epic, polished, slightly fantastical mood, rich contrast, deep shadows with vibrant accent colors, dynamic angle (low or high), atmospheric depth, No text, no watermark.
GPT 이미지 2로 생성한 앱 아이콘 보드

앱 아이콘 보드

GPT 이미지 2· Image
Design a premium app icon presentation board for a children's mobile game icon, inspired by Apple Human Interface Guidelines, Google Material Design documentation, and professional Behance case studies. Use a clean editorial layout with a minimalist white background, subtle light-gray construction grids, generous whitespace, and modern Swiss-inspired typography. The presentation should feel like it was created by a senior brand designer at a top mobile game studio.Place the final app icon as the hero element in the center, occupying approximately 60–70% of the page. Surround it with elegant technical annotations connected by thin gray leader lines. Include labels such as Safe Area, Adaptive Icon Mask, Optical Center, Visual Weight, Focal Point, Padding Zone, Keyline Grid, and Recommended Crop Area.Overlay a precise construction grid consisting of circles, squares, alignment guides, center lines, diagonal guides, proportional geometry, and adaptive icon boundaries. The grid should be subtle, using thin cyan or light-blue lines with low opacity, similar to Apple's icon construction documentation.Around the hero icon, include clean presentation sections with consistent spacing:Hero Icon ShowcaseConstruction GridSafe Area & Adaptive Icon MaskOptical AlignmentVisual Hierarchy DiagramColor Palette Swatches with HEX valuesTypography & Annotation StyleIcon Breakdown (Outline → Base Colors → Shading → Final Artwork)Small Size Preview (1024px, 512px, 256px, 128px, 64px, 48px)Platform Preview (Square, Rounded Square, Circle, Squircle)Use elegant monochrome labels, thin divider lines, subtle spacing, and professional alignment throughout the page. Typography should resemble modern design documentation with lightweight sans-serif fonts, small uppercase section titles, and minimal captions.The overall aesthetic should communicate precision, craftsmanship, and premium product design rather than marketing. Avoid gradients, decorative backgrounds, heavy shadows, colorful graphics, unnecessary icons, or visual clutter. The presentation should look like an official design specification created for a world-class mobile game studio and be suitable for Behance, Dribbble, or an AAA game design portfolio. Ultra-clean, highly organized, modern editorial design, perfect spacing, museum-quality layout, premium UI/UX presentation style.Avoid mockup devices, laptops, desks, hands, unnecessary 3D objects, colorful backgrounds, random decorative elements, excessive shadows, glassmorphism, neon effects, AI-style layouts, distorted grids, inconsistent spacing, low-resolution typography, cluttered composition, watermarks, logos, and marketing banners. Focus entirely on a clean professional icon specification sheet.
aacc56f2eccdceb35eb26a505aeb3232
GPT 이미지 2로 생성한 AI 이미지 예시

선글라스 문어 비즈니스 사진

GPT 이미지 2· Image
Hyperrealistic studio photograph of a majestic octopus with iridescent blue-grey mottled skin, wearing sleek minimal black sunglasses, positioned confidently on a glossy black obsidian pedestal in a seated, executive-like pose. Four tentacles each perform a distinct task simultaneously: one typing on a matte-black MacBook laptop, one holding a glowing smartphone displaying an upward revenue graph, one gripping a translucent holographic globe with pulsing connection nodes, one balancing a small stack of gold coins mid-transfer. Behind the octopus, five ultra-thin curved monitor screens arc in a 180-degree semi-circle, displaying SaaS dashboard interfaces, analytics charts, and a world map with glowing user-activity dots. Pure matte black studio background, no clutter, no extra props. Octopus skin rendered with ultra-fine photorealistic texture, subtle wet sheen on the suction cups. Shot on ARRI ALEXA 65 with a 50mm prime lens, shallow depth of field with the subject tack-sharp and the background screens falling into soft bokeh, dramatic single-source cool blue-white rim light from upper left, subtle warm red accent bounce light from lower right (hex ee4650), high-contrast chiaroscuro, glossy obsidian surface reflecting faint screen glow, ultra-detailed, 8k, commercial advertising photography --ar 4:5 --v 6.1 --style raw
2fd8674802b8a6aa04be8392eaa488d2
미드저니로 생성한 AI 이미지 예시

붉은 눈의 애니풍 얼굴 클로즈업

미드저니· Image
面容特写, 异色眼睛瞳孔内呈现红色,皮肤质感细腻,层次感油腻光,厚斜刘海,长直发,侧光透过镜头外缝隙在脸上形成独特唯美花纹阴影效果,轻质感纱巾缠绕脖子,游戏CG,电影质感,阴影描边,二次元,日系,梦幻,唯美,微风,动态模糊,明暗对比强烈,暗色系 --ar 9:16 --profile zrondri --hd
d0b7d9fda92227198bea07638a2e18cf
GPT 이미지 2로 생성한 졸업식 기념 사진 배경

졸업식 기념 사진 배경

GPT 이미지 2· Image
cinematic graduation celebration photo concept, graduation cap and diploma on campus lawn, warm sunlight, emotional festive mood, shallow depth of field, no text
미드저니로 생성한 포스터 이미지

편집 화면 속 눈 클로즈업

미드저니· Image
1. json prompt version ```json { "subject": { "description": "A cropped, low-resolution digital image of a person's eyes inside a rectangular selection box, resembling an image-editing canvas or screenshot. Only the eye area and eyebrows are visible; the rest of the face and body are cut off.", "mirror_rules": "Not a mirror selfie. Preserve the screen-capture / interface-editing feel rather than treating it as a conventional portrait.", "age": "adult or young adult, exact age unclear due to extreme crop and pixelation", "expression": { "eyes": { "look": "wide, direct, slightly tense gaze", "energy": "alert, uncanny, quietly confrontational", "direction": "looking straight toward the viewer" }, "mouth": { "position": "not visible", "energy": "not applicable" }, "overall": "fragmented and decontextualized, with the eyes carrying all emotional weight" }, "face": { "preserve_original": "Preserve the exact cropped eye-band framing, visible eyebrows, dark pupils, low-resolution halftone texture, and slight pink-gray screen tint.", "makeup": "No clear makeup is visible. The face should read as a raw, pixelated screen image rather than a polished beauty portrait." } }, "hair": { "color": "not visible", "style": "not visible", "effect": "not applicable" }, "body": { "frame": "not visible", "waist": "not visible", "chest": "not visible", "legs": "not visible", "skin": { "visible_areas": "Only a narrow strip of facial skin around the eyes, eyelids, under-eye area, bridge of the nose, and brow region is visible. No shoulders, chest, waist, hips, legs, back, or arms are visible.", "tone": "pale pink-gray skin tone distorted by screen pixels and scanline-like color separation", "texture": "The skin appears roughened by digital pixel structure, like it would feel visually grainy or screen-printed rather than physically smooth; any natural skin texture is obscured by the low-resolution display pattern.", "lighting_effect": "The skin is not lit by natural photographic light; it appears illuminated through a flat digital screen or screenshot, with no realistic shadow depth, only pixel-level contrast and faint magenta-gray banding." } }, "pose": { "position": "Only the eyes are framed horizontally inside a selected rectangular crop.", "base": "No body pose visible.", "overall": "The composition feels like a fragment being actively selected, edited, or dragged inside a design interface." }, "clothing": { "top": { "type": "not visible", "color": "not visible", "details": "not visible", "effect": "not applicable" }, "bottom": { "type": "not visible", "color": "not visible", "details": "not visible" } }, "accessories": { "device": "Implied computer interface or editing software canvas", "prop": "Multiple oversized black-and-white mouse cursor arrows surrounding and pointing toward the selected eye image" }, "photography": { "camera_style": "Not a traditional photograph; a screenshot-like digital collage or UI-capture aesthetic with a pasted low-resolution eye image.", "angle": "Straight-on, flat frontal view with no perspective depth.", "shot_type": "Extreme close-up crop of eyes embedded in a graphic interface composition.", "aspect_ratio": "Overall image is near 4:5 vertical canvas; the selected eye strip is a long horizontal rectangle.", "texture": "Visible pixel grid, moire-like vertical RGB scanlines, compressed low-res texture, hard-edged UI elements, clean white canvas background.", "lighting": "Flat screen illumination, no cinematic lighting, no natural directionality; contrast comes from the digital image and interface overlay.", "depth_of_field": "No optical depth of field; all elements appear flattened like a graphic screenshot." }, "background": { "setting": "Minimal white digital workspace or blank canvas", "wall_color": "white", "elements": [ "bright cyan-blue rectangular selection outline", "round blue resize handles on corners and midpoints", "small dark floating size tooltip near the top-right edge", "several repeated mouse cursor arrows arranged around the eye crop", "large empty white space above and below the selected image" ], "atmosphere": "Sparse, internet-art, meme-adjacent, slightly uncanny and obsessive", "lighting": "Pure digital white background with no physical environmental light" }, "the_vibe": { "energy": "hyper-focused, intrusive, glitchy", "mood": "uncanny, anxious, surveillance-like, absurdly intimate", "aesthetic": "early internet collage, screenshot art, low-res UI manipulation, digital found-image surrealism", "authenticity": "Should feel like an imperfect screenshot taken during editing, not a clean vector recreation or polished poster.", "intimacy": "The extreme eye crop creates uncomfortable closeness while the interface elements keep it emotionally detached.", "story": "It feels as if someone had been selecting a stranger's gaze on a blank canvas, and the repeated cursors are still converging on it.", "caption_energy": "something that was clicked too many times, a gaze trapped inside an editing box" }, "constraints": { "must_keep": [ "white blank background", "horizontal rectangular crop containing only both eyes and eyebrows", "cyan-blue selection box with circular resize handles", "visible pixelated / scanline texture across the eye image", "multiple large black-and-white cursor arrows pointing toward or around the eyes", "small size-tooltip near the upper right of the selection box", "flat screenshot-like composition", "uncanny direct gaze" ], "avoid": [ "do not turn it into a full-face portrait", "do not add visible body, clothing, hair, or environment", "do not make the image high-fashion or cinematic", "do not remove the UI selection handles", "do not smooth out the pixel grid", "do not replace the cursors with realistic hands or objects", "do not add dramatic shadows or realistic depth" ] }, "negative_prompt": [ "full body", "full face", "beauty portrait", "cinematic lighting", "realistic background", "smooth skin", "high resolution glamour", "fashion editorial", "hands", "extra facial features", "removed cursor arrows", "removed selection box", "3D depth", "polished vector art" ] } ``` 2. Natural language prompt Create a flat, screenshot-like digital collage on a mostly blank white canvas. In the center, place a long horizontal rectangular crop showing only a person's eyes, eyebrows, eyelids, under-eye area, and a narrow strip of pale pink-gray facial skin. The gaze should look straight at the viewer, wide and slightly tense, with an uncanny, alert emotional residue. Do not show the full face, mouth, hair, body, clothing, shoulders, chest, waist, hips, legs, arms, or any physical environment. The eye crop should look low-resolution and screen-captured, with visible pixel grid, vertical RGB scanline texture, faint moire, compression artifacts, and a slightly gray-magenta digital tint. The skin should not look polished or naturally lit; it should feel visually grainy, like a rough screen-printed or monitor-textured surface rather than smooth real skin. Lighting should be flat screen illumination with no realistic shadows, no cinematic directionality, and no depth-of-field blur. Surround the eye strip with a bright cyan-blue editing selection rectangle, including circular resize handles on the corners and midpoints. Add a small dark size-tooltip near the upper-right edge of the selected rectangle, as if the image is actively being resized in editing software. Place several oversized black-and-white mouse cursor arrows around the eye crop, some pointing toward the eyes and some overlapping the area below, creating a repetitive obsessive clicking energy. The composition should remain front-facing, flat, and graphic, with a near 4:5 vertical canvas and a large amount of empty white space above and below the selected eye strip. Preserve the imperfect screenshot feeling: slightly awkward placement, low-res texture, hard UI edges, and a casual internet-art quality. The image should feel as if someone had been selecting a stranger's gaze on a blank canvas, and the cursors are still converging on it. Avoid turning it into a full portrait, beauty shot, fashion editorial, cinematic scene, polished vector poster, or realistic environment. Keep the selection box, cursor arrows, pixelation, flat white background, and direct uncanny gaze as the core visual identity.
d1f13b0062b253274393e0ac28bf3555

시각적 서사의 완성 배경 디자인

배경은 모든 이미지의 분위기와 맥락을 결정짓는 핵심적인 요소입니다. 인물이나 제품이 돋보이려면 그에 걸맞은 조화로운 배경이 뒷받침되어야 합니다. AI 기술을 활용하면 물리적인 한계를 뛰어넘어 현실에 존재하지 않는 환상적인 풍경부터 초호화 인테리어, 정교한 스튜디오 세트까지 무엇이든 배경으로 구현할 수 있습니다. 캐럿은 여러분의 창작물이 최상의 퀄리티를 가질 수 있도록 목적에 맞는 다양한 배경 생성 프롬프트와 기법을 안내합니다.

목적별 최적의 배경 선택

이미지의 목적에 따라 어울리는 배경이 다릅니다. 제품 광고 이미지라면 '미니멀한 대리석 질감(minimalist marble texture)'이나 '세련된 파스텔 톤(sophisticated pastel tones)'이 제품을 돋보이게 합니다. 인물 포트레이트라면 '감성적인 카페 배경(atmospheric cafe)'이나 '도심의 보케 야경(city night bokeh)'이 인물의 분위기를 살려줍니다. 또한 유튜브 썸네일이나 웹사이트 메인 배너를 위해 '화려한 그래디언트(vibrant gradient)'나 '추상적인 3D 패턴(abstract 3d pattern)' 배경을 생성해 시각적 임팩트를 줄 수 있습니다.

조명과 깊이감을 살리는 묘사

단조로운 배경에서 벗어나 입체감을 주려면 조명과 원근감을 묘사하는 키워드가 필수입니다. '창문을 통해 들어오는 자연광(natural light through window)', '부드러운 스튜디오 조명(soft studio lighting)', '안개 낀 신비로운 분위기(misty and mysterious atmosphere)' 등을 추가해 보세요. 또한 '심도 깊은 원근감(deep perspective)'이나 '광각 렌즈 효과(wide angle view)'를 사용하면 공간이 훨씬 넓고 웅장해 보이는 효과를 얻을 수 있습니다. 배경에 배치된 사물들의 그림자 디테일까지 신경 쓴다면 더욱 실사적인 배경이 완성됩니다.

공간과 계절감의 변주

동일한 테마라도 공간과 계절에 따라 전혀 다른 느낌을 전달할 수 있습니다. 거실 배경이라면 '북유럽풍 인테리어(Scandinavian interior)'와 '모던 미니멀리즘(modern minimalism)' 중 선택할 수 있고, 자연 배경이라면 '눈 덮인 겨울 숲(snow-covered winter forest)'이나 '꽃이 만발한 봄 정원(blooming spring garden)' 등으로 계절감을 부여할 수 있습니다. AI 프롬프트의 강점은 단 몇 마디의 키워드 수정만으로 공간의 테마와 날씨, 시간대(낮, 밤, 노을)를 자유자재로 바꿀 수 있다는 점입니다.

고해상도 자산으로서의 배경 활용

AI로 생성한 배경은 화상 회의 배경화면, 합성용 소스, 게임 배경 일러스트 등 무궁무진하게 활용됩니다. 상업적인 품질을 확보하기 위해 'photorealistic', '8k resolution', 'highly detailed texture'와 같은 품질 강화 키워드를 잊지 마세요. 배경 이미지만 단독으로 생성한 후, 나중에 인물이나 제품을 합성하는 방식으로 작업 효율을 높일 수 있습니다. 캐럿의 배경 프롬프트 라이브러리를 통해 여러분의 프로젝트에 깊이와 전문성을 더할 완벽한 무대를 만들어 보시기 바랍니다.

Frequently Asked Questions

Related Prompts

폰 배경화면

폰 배경화면

13 prompts

인스타 사진

인스타 사진

15 prompts

드라마틱 조명

드라마틱 조명

89 prompts

제품 목업

제품 목업

230 prompts

Carat Prompt Gallery

Create trending AI prompts instantly

Prompts

  • Logo
  • PPT Design
  • Thumbnail
  • Poster
  • Web Novel Cover
  • ID Photo
  • Caricature
  • Product Detail Page
  • Banner

Models

  • GPT Image 2
  • Seedance 2.0
  • NanoBanana Pro
  • NanoBanana 2

Carat

  • Get started
  • About
  • Blog
  • Careers
  • Contact
  • Partnership
  • Terms
  • Privacy

© 2026 Paradot.Inc All rights reserved.