This ChatGPT photorealistic image prompt creates a candid phone photo of a drink spilled beside a commuter bicycle. Copy the full prompt or use the editable fields to change the drink, setting, crop, and light. A complete Korean version is included for readers who would rather edit the instructions in Korean.
The prompt suits blog posts, social content, and visual stories that need an unpolished everyday photo rather than a studio image. You can copy it as written or replace the drink, location, crop, and lighting fields listed below.
Can ChatGPT create a realistic smartphone photo from this prompt?
Yes. The first run included one tipped cup, a separate lid, strawberry soda, individual ice cubes, a blank paper sleeve, a cropped bicycle, and the photographer’s sneaker toe. The soft overcast light and awkward crop made the result feel closer to a quick phone snapshot than a commercial drink photo.
- Open ChatGPT on the web and select the image creation option shown in your interface.
- Copy the complete English prompt below.
- Send it without shortening the constraints on the first run.
- If you like the composition, continue in the same conversation when changing the light or weather.
OpenAI’s image generation guide recommends describing the image’s purpose, subject, setting, and visual style. It also recommends stating what should change and what should stay fixed when editing an image. Those instructions are separated into labeled blocks here so they are easier to modify.
Copy the complete ChatGPT photorealistic prompt
This is the exact text used for the first result. It was not reconstructed after the image was generated.
Create one image only. Use case: photorealistic-natural.
Asset type: vertical blog example image.
Primary request: a candid smartphone photograph taken seconds after an unbranded iced strawberry soda fell from the front basket of a commuter bicycle onto a narrow Korean city side-street sidewalk.
Scene/backdrop: ordinary gray interlocking concrete pavers beside a curb; only the lower front wheel, fork, and a small part of the matte charcoal bicycle basket enter from the upper-right edge.
Subject: one clear disposable cup lying on its side, its plain lid nearby, pale strawberry-red soda spreading in an irregular thin puddle, six to eight distinct ice cubes, tiny carbonation bubbles, one crumpled blank paper sleeve near the bicycle wheel. No people; only the cropped toe of a plain gray sneaker at the bottom edge suggests the photographer.
Style/medium: photorealistic raw smartphone documentary photo, an accidental everyday moment, not commercial food photography.
Composition/framing: exact 9:16 portrait, handheld waist-height view angled downward about 35 degrees, imperfect off-center crop, cup in the lower-middle, bicycle wheel partially cut by the frame, believable perspective.
Lighting/mood: soft overcast late-afternoon daylight, neutral color balance, low contrast, natural soft-edged shadows.
Materials/textures: porous concrete, scuffed clear plastic, watery translucent soda, melting ice, faint dirt in pavement joints, subtle phone-camera noise.
Constraints: physically plausible liquid spread and contact shadows; one cup only; no readable text; no brand marks; no logos; no watermark; no poster layout; no staged styling; no dramatic cinematic grading; no illustration; no CGI; no duplicated objects; no floating ice; no malformed cup or bicycle geometry.Bonus: copy the full Korean prompt
If Korean is easier for you to edit, copy the version below. I also sent it by itself in a fresh ChatGPT conversation to make sure the translated instructions could produce the intended scene.
이미지 한 장만 생성하세요. 사용 사례: 사실적인 자연스러운 사진.
에셋 유형: 세로형 블로그 예시 이미지.
주요 요청: 출퇴근용 자전거 앞바구니에서 무상표 아이스 딸기 탄산음료가 좁은 한국 도시 이면도로의 보도로 떨어진 지 몇 초 뒤를 포착한 자연스러운 스마트폰 사진.
장면과 배경: 연석 옆의 평범한 회색 맞물림 콘크리트 보도블록. 오른쪽 위 가장자리에는 자전거 앞바퀴의 아래쪽, 포크, 무광 짙은 회색 바구니의 작은 일부만 들어오게 하세요.
피사체: 투명한 일회용 컵 한 개가 옆으로 누워 있고, 아무 무늬 없는 뚜껑은 가까이에 놓여 있습니다. 옅은 딸기색 탄산음료가 불규칙하고 얇은 웅덩이로 퍼져 있으며, 서로 구분되는 얼음 여섯 개에서 여덟 개, 작은 탄산 기포, 자전거 바퀴 근처에 구겨진 무지 종이 슬리브 한 개가 있습니다. 사람은 없어야 합니다. 아래쪽 가장자리에서 잘린 평범한 회색 운동화 앞코만 사진을 찍은 사람의 존재를 암시합니다.
스타일과 매체: 사실적인 가공 전 스마트폰 다큐멘터리 사진. 상업용 음료 사진이 아니라 우연히 생긴 평범한 일상의 순간.
구도와 프레이밍: 정확한 9:16 세로 비율. 허리 높이에서 손으로 들고 약 35도 아래를 향해 촬영한 시점. 완벽하지 않은 비대칭 크롭. 컵은 아래쪽 중앙 부근에 놓고 자전거 바퀴는 프레임에 일부 잘리게 하며, 원근감은 현실적이어야 합니다.
조명과 분위기: 부드럽게 흐린 늦은 오후의 자연광, 중립적인 색 균형, 낮은 대비, 가장자리가 부드러운 자연스러운 그림자.
재질과 질감: 다공성 콘크리트, 긁힌 투명 플라스틱, 물처럼 묽고 반투명한 탄산음료, 녹고 있는 얼음, 보도블록 틈의 희미한 먼지, 은은한 휴대폰 카메라 노이즈.
제약 조건: 액체가 퍼진 모양과 접촉 그림자는 물리적으로 자연스러워야 합니다. 컵은 한 개만 있어야 합니다. 읽을 수 있는 글자, 브랜드 표시, 로고, 워터마크, 포스터형 레이아웃, 연출된 스타일링, 극적인 영화 색보정, 일러스트, CGI, 복제된 사물, 공중에 뜬 얼음, 변형된 컵이나 자전거 구조를 넣지 마세요.The Korean prompt produced the cup, lid, ice, paper sleeve, bicycle, and sneaker requested in the instructions. A fresh generation can still move the cup, ice, or basket. If you already have a composition you like, continue editing that image in its existing conversation.
Which parts of the prompt should you change?
Search for the exact phrase in the second column and replace one field at a time. If you change the drink, update its color and physical behavior in the subject and material lines too.
| Field | Find this text | Possible replacement | Check before sending |
|---|---|---|---|
| Drink | iced strawberry soda |
iced americano or orange juice |
Update the liquid color and remove carbonation if it no longer applies |
| Accident | fell from the front basket |
slipped from a paper shopping bag |
Remove or replace scene objects that no longer explain the accident |
| Location | Korean city side-street sidewalk |
quiet apartment walkway |
Match the pavement, curb, and background to the new setting |
| Aspect ratio | exact 9:16 portrait |
4:5 portrait or 1:1 square |
A new crop may reveal objects that were previously cut off |
| Camera angle | angled downward about 35 degrees |
angled downward about 55 degrees |
Keep the camera height fixed if you want to isolate the angle change |
| Time and weather | soft overcast late-afternoon daylight |
blue hour just after rain |
State which objects and positions must remain unchanged |
Example: change strawberry soda to iced coffee
Replace iced strawberry soda with iced americano. Then replace pale strawberry-red soda with dark translucent brown coffee and delete tiny carbonation bubbles. This coffee version is an editing example, not one of the three generated images in this article.
How do you keep the composition and change only the light?
The third image was made in the English image’s original conversation. The follow-up changed the dry late afternoon to blue hour after rain while explicitly preserving the crop, bicycle, cup, lid, spill, ice, sleeve, and shoe.
Edit this generated image as a controlled lighting-and-weather variation. Change only the environmental condition from a dry, softly overcast late afternoon to blue hour just after rain. Darken and lightly wet the pavement outside the existing soda spill. Add restrained warm amber storefront light reflecting across a few pavers and cool blue ambient light in the shadows. Add a small number of fine raindrops and subtle wet highlights on the existing bicycle wheel, basket, clear cup, lid, ice, and paper sleeve.
Keep unchanged: the exact 9:16 crop, camera position, downward viewing angle, curb and road geometry, interlocking-paver pattern, bicycle wheel, fork and basket position and proportions, cup shape and position, lid position, strawberry-red liquid spill shape and amount, number and position of ice cubes, paper sleeve, gray sneaker toe, all object scale, and documentary smartphone realism.
Constraints: physically plausible reflections and contact shadows; no new objects; no people; no text; no brand marks; no logos; no watermark; no cinematic fog; no heavy neon; no dramatic color grading; no illustration or CGI; do not move, duplicate, remove, or redesign any existing object.| Variable | Original | Edited result | Observed change |
|---|---|---|---|
| Time | Overcast late afternoon | Blue hour after rain | The image became darker and the shadows shifted cooler |
| Pavement | Dry outside the soda spill | Lightly wet sidewalk and road | Fine wet highlights appeared across the pavers |
| Light | Neutral diffuse daylight | Cool ambient light with amber reflections | The amber reflection was slightly stronger than requested |
| Fixed layout | Crop, camera, bicycle, cup, lid, spill, ice, sleeve, and shoe | The main positions and proportions remained stable | |
What made the image look less like an AI render?
The useful instructions were not generic quality labels such as 8K or masterpiece. The prompt gave the camera a reason to exist: someone had just dropped a drink and pulled out a phone. It also allowed an awkward crop, cutting the bicycle at the right edge and the shoe at the bottom.
Material words mattered too. The soda was described as watery translucent, the cup as scuffed clear plastic, and the sidewalk as porous concrete. The first result showed dirt in the joints and small surface marks instead of smoothing the setting into a clean advertisement.
What should you change when the cup or liquid looks wrong?
The English original passed on its first run, though the camera looked steeper than the requested 35 degrees and the basket was larger than expected. The Korean result also made the basket and wheel too prominent. To reduce that area, add the basket occupies less than 12 percent of the frame after the basket description. That correction has not been generated for this article, so it is a next-run suggestion rather than a verified result.
If the liquid looks like gel, avoid piling on more color words. Describe viscosity and depth with terms such as watery, translucent, and thin puddle. If an edit moves the cup or bicycle, repeat their positions under Keep unchanged and continue in the conversation containing the source image.
How was this prompt verified?
- Date: September 5, 2026 KST
- Tool: signed-in ChatGPT web
- Observed interface label:
이미지 만들기, which means image creation - Internal model name: not shown in the interface and therefore not claimed
- Image size: all three downloads were 941 by 1672 pixels
- Attempts: one English original, one controlled edit, and one independent Korean generation
- Rejected outputs: none
The topic was discovered through case 376 in awesome-gpt-image-2, which catalogues a spilled street drink photographed like a phone snapshot. The case’s prompt and image were not copied or used as generator input. The collection lists a post by @Shinning1010 as its original source.
Common questions
Will the same prompt generate the same image twice?
No identical result is guaranteed. The prompt narrows the scene, object count, materials, and framing, but each fresh generation can arrange those details differently. Continue editing an image you already like instead of starting over when composition matters.
Can I use the Korean prompt instead?
Yes. The Korean version above is ready to copy and produced the requested subject in a fresh conversation. Use the version you can edit most accurately. A new generation may arrange the smaller details differently in either language.
Can I change the lighting without moving the objects?
Use the original image’s conversation and separate the request into Change only and Keep unchanged instructions. The blue-hour variation preserved the major layout, although small ice edges and droplets were reinterpreted.
For the simplest first result, run the prompt unchanged. Then edit one field, such as the drink or time of day, in the same conversation.