← all repositories
helloianneo/ian-xiaohei-scenes

Turning Chinese Tech Stress Into Tiny Physical Dioramas

A Codex Skill that replaces generic Chinese article illustrations with physical dioramas starring a black mascot, real objects, and very little text.

ian-xiaohei-scenes
Velocity · 7d
+2.7
★ / day
star history

What it does

Ian Xiaohei Scenes is a Codex Skill—essentially a prompt-and-workflow package for an AI agent—that generates two kinds of visuals for Chinese-language content. Standard mode produces 16:9 images of a black mascot called Xiaohei interacting with a real-world object on a white background, meant to crystallize a single abstract idea like message overload or production alert. A second “easter egg” mode stitches five to eight such object moments into an ultra-wide scroll for project retrospectives or personal narratives. The goal is to turn a reader’s situation into a visible physical scene they recognize in under a second.

The interesting bit

The author treats illustration not as decoration but as translation: the skill first extracts the “real situation” from an article—work stress, a turning point, a project failure—then forces the AI to render it as a physical action with a real prop, not a metaphorical cloud or arrow diagram. It is deliberately anti-PPT; the documentation explicitly warns against flowcharts, numbered timelines, and cramming text into the frame.

Key highlights

  • Built specifically for Chinese content, with short Chinese labels as a core visual element.
  • Two distinct output modes: single-scene 16:9 article headers and ultra-wide “long scroll” story maps.
  • Heavy emphasis on constraints: pure white background, one core physical action per image, and the mascot must be doing the action—not just standing nearby.
  • Ships with seven “master” example images that serve as quality benchmarks rather than copy-paste templates.
  • Includes a full agent workflow and QA checklist for generation, review, and style drift control.

Caveats

  • The generated images are raster PNGs, not editable vectors, and the skill explicitly warns that AI image models often hallucinate or misspell Chinese text.
  • It is narrowly tuned for a specific visual identity (white studio, real objects, Xiaohei mascot); if your brand needs polished corporate flat design or app screenshots, this is the wrong tool.
  • The README notes that users should not mechanically replicate the example compositions, which suggests the output still requires human taste and curation.

Verdict

Worth a look if you publish Chinese-language project write-ups, tutorials, or indie-builder stories and want a consistent, lightly absurd visual voice. Skip it if you need editable diagrams, commercial key visuals, or illustrations in languages other than Chinese.

Frequently asked

What is helloianneo/ian-xiaohei-scenes?
A Codex Skill that replaces generic Chinese article illustrations with physical dioramas starring a black mascot, real objects, and very little text.
Is ian-xiaohei-scenes open source?
Yes — helloianneo/ian-xiaohei-scenes is open source, released under the MIT license.
How popular is ian-xiaohei-scenes?
helloianneo/ian-xiaohei-scenes has 519 stars on GitHub.
Where can I find ian-xiaohei-scenes?
helloianneo/ian-xiaohei-scenes is on GitHub at https://github.com/helloianneo/ian-xiaohei-scenes.

heatdrop uses Google Analytics to see which pages get read — nothing else. Your call. How we handle data.