InkRoom
InkRoom
Public prompt信息图解

信息图 / 教育视觉图 - Celica Real vs Miniature Comparison

✓Original text is publicUpdated:
GPT 8.0/10
信息图 / 教育视觉图 - Celica Real vs Miniature ComparisonPrompt preview
Results depend on the model, inputs and generation settings.

Full prompt

Original author text

This public page preserves the author's original prompt. Review the preview and any published evaluation, then copy the text or open it as a draft. Choose the model, references and current cost in the creation workspace before generating.

Goal: Create a vertical comparison image showing the same white Toyota Celica GT-Four in three stacked photographic panels, comparing three realism styles: a real car, a real car photographed to look like a miniature, and a plastic model photographed to look like a real car. Canvas: Portrait 3:4 composition, 768 × 1024 style, divided into exactly 3 equal horizontal panels separated by thin white divider lines. No captions or overlay text outside the car details. Subject: A {argument name="car model" default="white Toyota Celica GT-Four ST205"} shown from a front-left three-quarter angle, low stance, white multi-spoke wheels, large rear wing, hood vent, round twin headlamps, front bumper fog lights, side decal reading “GT-FOUR,” and a Japanese license plate with the visible number {argument name="license plate number" default="22-31"}. The car should remain visually consistent across all three panels. Setting: Nighttime urban industrial road scene in {argument name="location style" default="Japan"}, wet asphalt after rain, reflective puddles, concrete pillars, chain-link fencing, bridge or elevated roadway structures, and warm streetlights receding into the background with bokeh. Panel 1: Top panel shows the real full-size car photographed naturally. Sharp, realistic automotive photography, crisp body details, accurate proportions, deep reflections on wet pavement, moderate background blur, cinematic night lighting. Panel 2: Middle panel shows the same real car but photographed to look miniature. Use tilt-shift depth of field, stronger blur at foreground and background, slightly toy-like scale illusion, softened details, shiny wet road reflections, and a more diorama-like feel while still using the same viewpoint. Panel 3: Bottom panel shows a plastic scale model of the same car photographed to look like a real full-size car. Keep the car highly realistic but subtly model-like in the surface finish and fine details; use shallow depth of field, realistic reflections, convincing low-angle automotive composition, and the same wet night street environment. Visual style: High-quality photorealistic comparison, cinematic contrast, glossy white paint, warm orange and cool blue night tones, realistic lens bokeh, no people, no extra vehicles, no logos or watermarks. Maintain exactly 3 panels and exactly 1 car per panel.

Adjust the model, settings and references before generating.

Use prompt →

Prefill first, then confirm generation and the current cost.

GPT evaluation

Overall score
8.0 / 10

GPT-6-luna

AI evaluation for creative reference

Model: GPT-6-luna Overall score: 8 / 10 Prompt quality: 8.7 / 10 Image quality: 8.5 / 10 Prompt–image alignment: 7 / 10 画面质感与三段式构图出色,但中、下两格与真实汽车的差异不够醒目,未充分呈现微缩模型对比,建议优化后上线。

Prompt quality8.7 / 108.7
Image quality8.5 / 108.5
Prompt–image alignment7.0 / 107.0

Community reviews

No creator reviews yet.

View and review →