> For the complete documentation index, see [llms.txt](https://docs.qolaba.ai/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.qolaba.ai/image-generation/image-editing/image-variation.md).

# Image Variation

Image Variation generates new images inspired by one or more reference images — producing aesthetically diverse versions that maintain the core subject and overall composition of the original while exploring different visual interpretations.

***

#### What Image Variation Does

Image Variation takes your uploaded reference images and generates alternative versions that capture the essence of the original — same subject, similar composition — but with natural variation in style, detail, color, and rendering. It is powered exclusively by **Nano Banana 2**, Qolaba's flagship Gemini model.

This is distinct from image-to-image generation, which uses a prompt to explicitly direct a transformation. Image Variation produces organic creative alternatives without requiring detailed transformation instructions.

***

#### Image Variation vs. Image-to-Image

|                      | Image Variation                          | Image-to-Image                          |
| -------------------- | ---------------------------------------- | --------------------------------------- |
| **Model**            | Nano Banana 2 only                       | Multiple models available               |
| **Reference images** | Up to 13                                 | Varies by model (up to 15)              |
| **Prompt required**  | Optional                                 | Recommended for direction               |
| **Output**           | Creative variations of the reference     | Directed transformation based on prompt |
| **Best for**         | Style exploration, creative alternatives | Specific, guided transformations        |

Use Image Variation when you want to explore what the image could look like with natural creative diversity. Use image-to-image when you have a specific transformation in mind and need to direct it with a prompt.

***

#### How to Use Image Variation

* **Step 1 —** Open **Image Editing → Image Variation** from the workspace.
* **Step 2 —** Upload up to **13 reference images**, or select from your generation history. The more references you provide, the more the model understands the visual language you want to explore.
* **Step 3 —** Optionally add a prompt to loosely guide the variation direction — style, mood, or specific elements to emphasize.
* **Step 4 —** Configure output settings — quality and dimensions.
* **Step 5 —** Review the credit cost and click **Generate**.

***

#### When to Use Image Variation

* **Creative exploration** — when you want to see alternative interpretations of an existing image without specifying exact changes
* **A/B testing visuals** — generate several variations of a marketing creative and select the strongest for production
* **Design iteration** — explore how an image looks across different stylistic interpretations before committing to a direction
* **Style consistency testing** — upload multiple brand assets as references and generate variations to see how well the model captures the brand's visual identity
* **Client options** — produce a range of alternatives from a confirmed base image to present multiple directions without full regeneration

***

#### Tips for Effective Image Variation

* **Upload multiple references for richer variation** — a single reference produces limited variation range. Uploading 3–5 references that share a consistent visual style gives the model a broader understanding of the aesthetic you want to explore
* **Use clean, high-quality references** — the quality of the variation output reflects the quality of the input references
* **Add a loose prompt for directional control** — if you want variations to lean toward a specific style or mood without fully directing a transformation, a brief style prompt (*"warmer color palette"*, *"more dramatic lighting"*) nudges the output without over-constraining it
* **Generate multiple variations at once** — set generations to 3 or 4 to produce a range of alternatives in one run, then select the strongest for further refinement or production


---

# Agent Instructions
This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com.

## Querying This Documentation
If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter:

```
GET https://docs.qolaba.ai/image-generation/image-editing/image-variation.md?ask=<question>
```

The question should be specific, self-contained, and written in natural language.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.