GPT Image 2 Webinar Thumbnail Prompt – Clean Japanese Key Visual

About this prompt

Copy this webinar thumbnail prompt prompt and run it in GPT Image 2 exactly as written. The wording asks the model for poster work in a tech and story context and keeps the full description that produced the original gallery result, so you can study a working example instead of guessing at phrasing from scratch. Iterate one clause at a time and regenerate to compare.

To reproduce the result faithfully, keep the template intact and change only subject-specific details such as names, colors, and scene elements first. The prompt is written in English, so keep any edits you make in the same language to preserve the model's typography and tone. Composition, lighting, and texture clauses carry most of the final look, so adjust the aspect ratio or quality settings only after the core wording already matches what you want.

This prompt is adapted from the awesome-gpt-image-2 collection under the MIT license and was originally shared by @kawai_design. The original case is titled “综合应用场景图” in the upstream gallery. The wording here matches the original text so credit and reproducibility stay intact, and the upstream entry documents the author's own notes.