信息图与卡片

百物场景清单图

把一个地点或主题变成一幅信息密集的插画场景:画面内恰好包含 100 个命名元素,并附内嵌清单——适合找找看海报和展示细节功力的作品。

更新日期:

完整提示词

英文提示词包含建议的示例需求。生成前请修改主体和细节。原始示例输入和生成参数尚未完全核实。

推荐模型: GPT Image 2

Compose a single rich, coherent illustrated scene that contains exactly one hundred distinct, findable elements, and pair it with a built-in inventory that names all one hundred inside the image — either a clean numbered legend panel along one side or the bottom, or small numbered tags in the scene keyed to a compact list. Build the setting described in the user request and populate it with believable clusters of objects, creatures, furnishings, and details, each rendered large enough to be spotted and identified. Keep one consistent, detailed illustration style with a warm cohesive palette and soft global lighting so the busy composition still feels unified rather than chaotic. Make the legend typography small, tidy, and correctly spelled, matched one-to-one with the scene: every listed item must actually appear, no item may be listed twice, and nothing prominent in the scene should be missing from the list. The result should reward slow looking, like a seek-and-find poster crossed with an illustrated inventory sheet. No watermarks, no brand names or logos, no duplicated entries. User request: a cozy mountain cabin in winter, or a chef's kitchen during dinner rush

官方示例

百物场景清单图
示例模型: Seedream 5.0 Lite
百物场景清单图
示例模型: Seedream 5.0 Lite

来源与署名

由 OmniArt 改编和重写。以下列出来源署名、修改说明及使用条件。

  • 100 Elements Scene Inventory

    ZeroLu/awesome-gpt-image

    README states CC BY 4.0 while the current LICENSE is MIT; attribution and the umbrella MIT notice are retained in THIRD_PARTY_NOTICES.md.

    Rewritten and consolidated into an OmniArt recipe archetype; template tokens removed; named people, brands, platforms, games, and film IP generalized; no raw prompt text or source image copied.

来源与署名

相关提示词配方