# What Scene Tags Can I Use for Image Generation in Awesome‑GPT‑Image‑2?

> Explore available scene tags for Awesome-GPT-Image-2 image generation. Discover creative, tech, commerce, education, social, fashion, food, travel, story, and history tags to guide AI.

- Repository: [苍何/awesome-gpt-image-2](https://github.com/freestylefly/awesome-gpt-image-2)
- Tags: how-to-guide
- Published: 2026-08-28

---

**Awesome‑GPT‑Image‑2 provides ten predefined scene tags—`creative`, `tech`, `commerce`, `education`, `social`, `fashion`, `food`, `travel`, `story`, and `history`—defined in [`data/style-library.json`](https://github.com/freestylefly/awesome-gpt-image-2/blob/main/data/style-library.json) to categorize visual context and guide AI image generation.**

The scene tag system in Awesome‑GPT‑Image‑2 allows developers and prompt engineers to classify image generation requests by narrative context and functional purpose. These tags act as metadata filters that help the style library select appropriate templates and guide the generation engine toward contextually relevant compositions.

## Where Scene Tags Are Defined

The canonical list of scene tags lives in **[`data/style-library.json`](https://github.com/freestylefly/awesome-gpt-image-2/blob/main/data/style-library.json)** within the **scenes** array. Each entry contains an identifier (`id`), a short label (`value`), and localized titles for multilingual support.

According to the Awesome‑GPT‑Image‑2 source code, the scenes array maps functional categories to prompt templates, enabling the system to filter cases by narrative context or suggest appropriate styling for specific domains.

## Complete List of Available Scene Tags

The repository currently exposes ten distinct scene tags. Each tag includes both English and Chinese localized titles to support international workflows:

| Tag ID | Label | English Title | Chinese Title |
|--------|-------|---------------|---------------|
| `creative` | Creative | Creative | 创意 |
| `tech` | Tech | Tech | 科技 |
| `commerce` | Commerce | Commerce | 商业 |
| `education` | Education | Education | 教育 |
| `social` | Social | Social | 社媒 |
| `fashion` | Fashion | Fashion | 时尚 |
| `food` | Food | Food | 食品饮品 |
| `travel` | Travel | Travel | 旅行 |
| `story` | Story | Story | 叙事 |
| `history` | History | History | 历史 |

Selecting `travel` signals the engine to favor city maps, street views, or landscape compositions, while `fashion` triggers portrait-oriented templates and style references.

## How to Use Scene Tags in Code

Scene tags integrate directly into the **gpt-image-2-style-library** skill and API requests. You pass the `scene` parameter alongside `style` and additional tags to contextualize generation.

### Style Library Skill Integration

When calling the style library skill, specify the scene to filter template selection:

```js
// Requesting a travel-scene infographic
await gptImage2StyleLibrary({
  style: "infographic",
  scene: "travel",          // scene tag identifier
  tags: ["Map", "Travel"],
});

```

```js
// Generating a fashion-scene portrait
await gptImage2StyleLibrary({
  style: "photography",
  scene: "fashion",         // scene tag identifier
  tags: ["Portrait", "Fashion"],
});

```

### API Query Filtering

You can also filter existing case galleries by scene tag via the REST API:

```js
// Filtering the gallery for education-scene cases
fetch("https://gpt-image2.canghe.ai/api/cases?scene=education")
  .then(r => r.json())
  .then(cases => console.log(cases));

```

## Technical Implementation Details

The scene tag system spans three critical files in the repository:

- **[`data/style-library.json`](https://github.com/freestylefly/awesome-gpt-image-2/blob/main/data/style-library.json)** – Stores the master **scenes** array containing all tag definitions, localized strings, and metadata mappings.
- **[`agents/skills/gpt-image-2-style-library/SKILL.md`](https://github.com/freestylefly/awesome-gpt-image-2/blob/main/agents/skills/gpt-image-2-style-library/SKILL.md)** – Documents the agent skill contract that consumes scene tags to route prompts to appropriate generation pipelines.
- **[`docs/templates.md`](https://github.com/freestylefly/awesome-gpt-image-2/blob/main/docs/templates.md)** – Contains full prompt template specifications where scene tags trigger context-specific styling instructions and composition guides.

When the skill processes a request, it cross-references the provided scene tag against the style library to retrieve predefined aesthetic parameters and negative prompts associated with that narrative context.

## Summary

- Awesome‑GPT‑Image‑2 defines **10 fixed scene tags** in [`data/style-library.json`](https://github.com/freestylefly/awesome-gpt-image-2/blob/main/data/style-library.json): creative, tech, commerce, education, social, fashion, food, travel, story, and history.
- Each tag includes bilingual support (English and Chinese) via the `value` and localized title fields.
- Pass scene tags via the `scene` parameter in `gptImage2StyleLibrary()` calls or as query parameters in API requests.
- The **gpt-image-2-style-library** skill uses these tags to select contextually appropriate prompt templates and styling guides.

## Frequently Asked Questions

### How do I add a custom scene tag to Awesome‑GPT‑Image‑2?

You cannot add custom scene tags without modifying the source. The repository uses a fixed enumeration defined in [`data/style-library.json`](https://github.com/freestylefly/awesome-gpt-image-2/blob/main/data/style-library.json). To extend the system, submit a pull request updating the **scenes** array in that file and regenerate the style library cache.

### Do scene tags affect the image generation model directly?

No, scene tags are metadata filters used by the style library skill to select templates and guide prompt construction. They do not pass through as raw tokens to the diffusion model; instead, they trigger pre-configured prompt templates and negative prompts defined in [`docs/templates.md`](https://github.com/freestylefly/awesome-gpt-image-2/blob/main/docs/templates.md).

### Can I use multiple scene tags in a single request?

The current implementation in [`SKILL.md`](https://github.com/freestylefly/awesome-gpt-image-2/blob/main/SKILL.md) suggests single-scene selection. Requests typically accept one `scene` string value. For hybrid contexts, use the `tags` array to supply additional descriptive keywords while keeping the primary `scene` as the dominant narrative category.

### What is the difference between `scene` and `tags` parameters?

The `scene` parameter accepts one of the 10 predefined categorical tags (creative, tech, etc.) that drive template selection. The `tags` array accepts arbitrary descriptive strings that refine the subject matter (e.g., "Portrait", "Map") but do not trigger specific style library routing logic.