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

# Image Editing

Refine your generated images with precise text instructions.

On **Ultimate**, you can also edit your persona photos directly in Studio: open a photo’s **More actions → Edit photo**. Follow [Edit an existing persona photo](/create/persona-photos.md#edit-an-existing-persona-photo) for the steps and how to make the result your current reference.

### What is Image Editing?

Image editing allows you to modify an existing generated image using a text prompt. Instead of generating a completely new image from scratch, you can keep the parts you like and change specific details.

{% hint style="info" %}
Edits cost **50 moments** per edit.
{% endhint %}

### How to Edit an Image

1. **Select an Image:** Choose which image on the generate page.
2. **Click Edit:** Click the **Edit** (wand icon) button.
3. **Enter Instructions:** Describe exactly what you want to change in the text box.
4. **Generate:** Click **Generate Edit** to create a new version of the image.

<figure><img src="https://3286776735-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2F9ACKOAU4nFWSXPeE2B4s%2Fuploads%2FtD6IVtgDGPyjsnorgT14%2Fimage.png?alt=media&amp;token=f3966b25-598c-4003-9600-60c1db92d315" alt="" width="375"><figcaption></figcaption></figure>

### Prompting Guide

For the best results, your instructions should be clear, unambiguous, and focused on the specific change.

#### 1. Be Clear and Specific

Name the exact element you want to change and state the operation clearly. Avoid vague pronouns like "it" or "that one."

* **Do:** "Add matching silver earrings and a necklace."
* **Do:** "Remove the hat."
* **Do:** "Replace the bread with a croissant."
* **Avoid:** "Change that thing to something else."

<figure><img src="https://3286776735-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2F9ACKOAU4nFWSXPeE2B4s%2Fuploads%2F4RTnVBVSoXoY7uHPUj7l%2Fimage.png?alt=media&amp;token=46750189-73a3-4e05-a73e-b68b102a07a1" alt="" width="375"><figcaption></figcaption></figure>

#### 2. "Lock" What Should Stay the Same

To prevent accidental changes to the rest of the image, explicitly state what should remain untouched.

* "Keep the pose unchanged."
* "Keep facial expression unchanged."
* "Keep the background unchanged."
* "Preserve the original lighting and composition."

<figure><img src="https://3286776735-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2F9ACKOAU4nFWSXPeE2B4s%2Fuploads%2FhVP2e5fFVgqiYxNptTPH%2Fimage.png?alt=media&amp;token=84fd16bf-f756-4539-8158-58e458b1c563" alt="" width="375"><figcaption></figcaption></figure>

#### 3. Use a Structured Format

A simple structure often works best: **\[Operation] + \[Target] + \[Change Details] + \[Constraints]**

* **Example:** "Replace the red dress with a blue summer dress, matching the fabric style, keeping her pose and the beach background unchanged."

{% columns %}
{% column %}

<figure><img src="https://3286776735-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2F9ACKOAU4nFWSXPeE2B4s%2Fuploads%2FZiQD1JrlLERYRPWySeZ3%2Fimage.png?alt=media&amp;token=0a7828d2-d20d-4f01-a81e-5c6ad3c46447" alt=""><figcaption></figcaption></figure>
{% endcolumn %}

{% column %}

<figure><img src="https://3286776735-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2F9ACKOAU4nFWSXPeE2B4s%2Fuploads%2FRVa8uNyobUtxKWpC20sE%2Fimage.png?alt=media&amp;token=3d37893e-0d6e-4508-85a2-fe1c44f389e6" alt=""><figcaption></figcaption></figure>
{% endcolumn %}
{% endcolumns %}

<figure><img src="https://3286776735-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2F9ACKOAU4nFWSXPeE2B4s%2Fuploads%2F0KNoCY12gOgAEdgLCqEX%2Fimage.png?alt=media&amp;token=4fce21eb-852a-4278-808a-defd4a9965f4" alt="" width="375"><figcaption></figcaption></figure>

#### 4. Common Operations & Templates

**Add an Item**

* Add **sunglasses** to **the woman**, matching **the modern style**, consistent with existing lighting. Keep **her expression and the background** unchanged.

**Remove an Item**

* Remove **the hat** from the image. Fill the area naturally to match the surrounding **hair texture**. Keep everything else unchanged.

**Replace an Object**

* Replace **the coffee cup** with **a wine glass**, keeping her **hand position** unchanged. Match the original **evening lighting**.

**Modify Attributes (Color/Material)**

* Change **the dress** to **red silk**. Keep **the jewelry and shoes** unchanged.

### Frequently Asked Questions

**Q: Can I edit an image multiple times?**

* Yes. You can edit an edited image again to refine details or add more changes. Each edit costs 50 moments.

**Q: Why did my edit change the whole image?**

* If the edit changed too much, try being more specific about what *not* to change. Adding "Keep the background and pose unchanged" to your prompt helps "lock" those elements.

**Q: Can I use editing to change the character's identity?**

* Sometimes, but results vary.
* Editing works best for outfits, backgrounds, and small details.
* If you want a different person, start with a fresh generation.

**Q: My edit was stopped by a safety check. What now?**

* Edit instructions and edited results go through the same safety checks as image generation. The notice names what was flagged, for example a real person's likeness or an instruction the safety check couldn't approve, so you can adjust your wording. Edits stopped before generation aren't charged. See [Safety Check Messages](/create/image-generation.md#safety-check-messages).

***

Next steps: See [Image Generation](/create/image-generation.md) and [Moments & Payments](/extra/moments-and-payments.md).