> 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-generation.md).

# Image Generation

Learn how to request and create stunning images with your AI companions using our image generation feature.

### Jump to

* [The Generation Studio](#the-generation-studio)
* [Persona photos (Ultimate)](/create/persona-photos.md)
* [How to request images](#how-to-request-images)
* [How chat images use reference photos](#how-chat-images-use-reference-photos)
* [Image generation guide](#image-generation-guide)
* [Advanced image generation mode](#advanced-image-generation-mode)
* [Reference photos](#reference-photos)
* [Image generation presets](#image-generation-presets)
* [Batch image generation](#batch-image-generation)
* [Image generation examples](#image-generation-examples)
* [Group image generation](#group-image-generation)
* [Editing images](#editing-images)
* [Troubleshooting](#troubleshooting-image-generation)
* [Frequently asked questions](#frequently-asked-questions)

### The Generation Studio

The Generate page is a redesigned **Generation Studio** — one place to create both **images and videos**:

* **Image / Video toggle** — Switch between generating images and videos right on the page. Videos no longer open in a separate window; see [Video Generation](/create/video-generation.md).
* **Resizable layout** — On desktop, drag the divider between the generation form and your gallery to rebalance the two. On mobile, your gallery opens as a full-screen sheet.
* **Unified gallery actions** — Favorite, highlight, publish, and tag your generations directly on each card. Card actions work the same across the Generate, Collection, profile, and character pages, and work on touch devices — no hover needed.
* **Filters and sorting** — Filter your gallery by companion and tags, and sort by recent or favorites.

{% hint style="info" %}
**Personas and persona photos are exclusive to Ultimate.** All Ultimate subscribers can generate photos of just their persona, include themselves in Studio group photos, and use their active persona in chat photos. Follow [Persona Photos](/create/persona-photos.md) for setup, choosing a reference, and costs.
{% endhint %}

### How to Request Images

{% hint style="info" %}
A standard single-companion chat image costs **50 Moments**. A request that includes your persona photo costs **60 Moments**. Group photos and higher Studio resolutions have different costs; see [group pricing](#group-image-pricing) and [Persona Photos](/create/persona-photos.md#moments-and-costs).
{% endhint %}

#### Basic Image Requests

Simply ask your companion to send you an image. Your companion will generate an image based on the conversation context and send it to you.

<figure><img src="https://3286776735-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2F9ACKOAU4nFWSXPeE2B4s%2Fuploads%2FWCc0nODazi93KcdjSvLN%2Fimage.png?alt=media&amp;token=3ff3114a-02f7-41ee-a183-3012a49053b0" alt=""><figcaption></figcaption></figure>

#### Instant Images

The instant image button generates an image based on your conversation context.

<figure><img src="https://3286776735-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2F9ACKOAU4nFWSXPeE2B4s%2Fuploads%2FJUS47EsnpwgLZL53HMHx%2Fimage.png?alt=media&amp;token=a267be4d-daa6-42de-8c9a-d5ff5dd272e9" alt="" width="93"><figcaption></figcaption></figure>

<figure><img src="https://3286776735-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2F9ACKOAU4nFWSXPeE2B4s%2Fuploads%2FzjrUxwMAkY2oekNHYX6Z%2Fimage.png?alt=media&amp;token=0a689f30-6505-4475-8d46-77de971c3139" alt=""><figcaption></figcaption></figure>

#### How Chat Images Use Reference Photos

When a chat image uses a **reference photo**, that photo guides the companion's physical appearance. This applies both when you ask for a photo in conversation and when you use the **instant image button**, including reference-based group photos.

Use your request to describe **the scene**: clothing, pose, activity, setting, lighting, and mood. You do not need to repeat the referenced companion's hair color, facial features, skin tone, or body type. Conflicting appearance descriptions in the conversation or an earlier image prompt should not be used to redesign that companion.

For example:

> Send a photo of yourself in a blue raincoat beside the lighthouse, holding a folded map in soft morning light.

The reference supplies the companion's appearance; the request supplies the raincoat, map, setting, and lighting. Asking for a different hair color or face in chat is not the way to change a referenced companion's identity.

If the photo includes an additional person **without a reference**, you can describe that person's appearance. Name who each detail belongs to so the request is clear. On Ultimate, if you use a [persona photo](/create/persona-photos.md) as a reference for yourself, that photo guides your appearance too. Chat images use the conversation's **Active persona**; selecting **None** does not fall back to your account default.

For your own custom character, choose an appropriate [Advanced Image Reference](/create/custom-characters.md#advanced-image-reference-optional) in the Character Studio when you want to change the visual source. A reference helps maintain consistency, but generated images can still vary.

This guidance is for images created from chat. For prompts you write directly on the Generate page, follow the [Advanced mode prompt tips](#advanced-mode-prompt-tips); for changes to an existing image, see [Image Editing](/create/image-editing.md).

### Advanced Image Generation Mode

{% hint style="info" %}
Advanced Image Generation now works for **every companion** — nothing to enable, no setup, no requirements. Just pick a companion and generate.
{% endhint %}

Advanced mode is the most powerful way to generate images, with the highest realism and prompt adherence. Images generated in Advanced mode more accurately preserve your companion's exact facial features, hair, skin tone, and body type.

{% hint style="success" %}
**Major upgrade (July 2026):** Advanced mode now runs on a new, far more capable image model. Results are sharper and more realistic, follow your prompts more faithfully, and Advanced is now **very capable of generating NSFW and intimate scenes** — a previous weak spot that's gone.
{% endhint %}

#### Reference Photos

You can steer any Advanced generation with up to **5 reference photos** pulled from your **entire gallery** — filter by companion, tags, favorites, or most recent to find the right ones. The model uses your references to guide likeness, style, and detail.

* **First reference included** — Your first reference photo is included in the base price. Each **additional** reference costs **10 moments**.
* **Picks replace the default photo** — In solo mode, your picked references replace the companion's own photo as the visual source, so your first pick is the included reference.
* **Completely optional** — Without any picks, the companion's own photo is used automatically as the reference
* **Great for consistency** — Reference a favorite previous generation to keep a look, outfit, or setting consistent across images

#### Advanced Mode for Custom Characters

Advanced mode works with every custom character out of the box — yours and other users' — with no setup step. By default the character's **avatar** is used as the generation reference, so for the best results give your character an avatar that shows their **full body** (you can change it any time in the Character Studio's Gallery tab).

For your own characters, you can optionally set a dedicated **reference photo** that Advanced generation uses instead of the avatar — pick any image you've generated of that character in the Character Studio's **Appearance** tab (free, and revertible from the History tab). The reference decides which features stay consistent across generations: whatever it shows is preserved, so pick one that shows off what you care about — including an **NSFW reference** if you want intimate features maintained in NSFW images. See [Custom Characters](/create/custom-characters.md#advanced-image-reference-optional) for details and best practices.

#### Premium vs. Advanced Mode

| Feature            | Premium Mode         | Advanced Mode                                                     |
| ------------------ | -------------------- | ----------------------------------------------------------------- |
| **Prompt style**   | Comma-separated tags | Natural scene descriptions                                        |
| **Prompt length**  | Up to 1,000 chars    | Up to 2,000 chars                                                 |
| **Accuracy**       | Good                 | Significantly higher — preserves exact likeness                   |
| **Realism**        | Good                 | Highest — sharper, more photorealistic results                    |
| **NSFW scenes**    | Excellent            | Very capable — handles bold and intimate scenes with ease         |
| **Presets & tips** | Available            | Available — combine presets with your own scene description       |
| **References**     | Not supported        | Up to 5 reference photos (first included, +10 moments each extra) |
| **Resolution**     | Standard             | 1K (Fast), 1.5K, or 2K (4K coming soon)                           |
| **Availability**   | All companions¹      | All companions                                                    |

¹ **V2 companions** (marked with the iridescent **V2** badge) generate exclusively in Advanced mode — Premium isn't offered for them.

#### How to Use Advanced Mode

1. **Open the Generate page** and select **Images**
2. **Choose Advanced mode** from the mode selector
3. **Add reference photos** (optional, up to 5) to steer the result — otherwise the companion's own photo is used automatically
4. **Write a natural scene description** (up to **2,000 characters**) — describe the setting, clothing, pose, lighting, and mood in plain language
5. **Select your aspect ratio**: 1:1, 16:9, 9:16, 4:3, or 3:4
6. **Choose your resolution**: 1K (Fast), 1.5K, or 2K (4K is coming soon). 1K and 1.5K cost the same, and 1K renders fastest
7. **Generate** your image

#### Advanced Mode Prompt Tips

Write your prompt as a natural scene description rather than comma-separated tags:

✅ **Good**: "Standing on a rooftop at sunset wearing a red dress, wind blowing through hair, city skyline in the background"

❌ **Bad**: "rooftop, sunset, red dress, wind, hair, city skyline, background"

Be descriptive about the scene you want — the more detail you provide about the setting, pose, and atmosphere, the better your results will be.

#### Advanced Mode Pricing

| Resolution    | Cost per companion |
| ------------- | ------------------ |
| **1K (Fast)** | 50 moments         |
| **1.5K**      | 50 moments         |
| **2K**        | 100 moments        |
| **4K**        | Coming soon        |

**1K (Fast)** uses the model's smallest output size, so it renders the quickest, and it costs the same as 1.5K. Pick **1.5K** for more pixels at no extra cost, or **2K** for the sharpest output at double the price.

Each **reference photo past the first** adds **10 moments** on top of the base price. The first reference (your companion's own photo, or your first pick) is always included.

***

{% hint style="info" %}
**Tip:** On the Generate page, you can click the **Reuse Prompt** button on any previously generated image to copy its prompt into the editor — great for iterating on a result you liked.
{% endhint %}

### Image Generation Presets

{% hint style="info" %}
Image generation presets now work in **both Premium and Advanced modes**.
{% endhint %}

Presets give you a quick way to generate great images without writing a prompt from scratch. Choose from curated preset categories and let the AI handle the rest.

#### Preset Categories

* **Outfits** — Pre-defined clothing and styling options
* **Actions** — Poses and activities for your companion
* **Scenes** — Settings and environments for your images

#### How to Use Presets

1. **Open the Generate page** and select **Images**
2. **Browse available presets** by category (shown above the prompt in both modes)
3. **Select a preset** to apply it — you can combine presets with your own prompt, or generate from presets alone
4. **Generate** your image

#### Availability

* Currently available for **female companions** (realistic + anime)
* Male companion presets are coming soon

***

### Batch Image Generation

{% hint style="info" %}
Companion batches are available in **Premium** and **Advanced solo** mode. Advanced generation with picked gallery references is single-image. On Ultimate, [solo persona photos](/create/persona-photos.md#generate-photos-of-just-you) have their own **1, 2, or 4** photo selector.
{% endhint %}

Generate multiple images at once instead of one at a time. Batch generation lets you create up to **8 images** in a single generation from the Generate page.

#### How to Use Batch Generation

1. **Open the Generate page** and select **Images**
2. **Choose Premium or Advanced solo mode** from the mode selector
3. **Set your prompt** and image settings
4. **Select the number of images** to generate (up to 8)
5. **Generate** — all images are created in one batch

Each Premium image, or Advanced image at 1K/1.5K, costs **50 Moments**, so 8 images cost **400 Moments**. Advanced at 2K costs **100 Moments per image**, or **800 Moments** for 8. Check the total before generating.

***

### Image Generation Guide

#### Comma-Separated Tags Format

Image prompts use **comma-separated tags** rather than full sentences:

✅ **Good Format**: modern outdoor pool, night time, dark green skinny bikini, standing pose by poolside, ambient blue lighting, elegant evening atmosphere, modern villa background, subtle city lights, cinematic composition

❌ **Bad Format**: I'm at a modern outdoor pool at night time wearing a dark green skinny bikini and standing by the poolside with ambient blue lighting

#### Key Principles

**1. Use Comma-Separated Tags**

* Keep tags 1-4 words each
* Put the most important tags first
* Add lighting, mood, and composition when it matters

### Image Generation Examples

#### Example 1: Casual Beach Scene

**Good Prompt**: beach setting, daytime, blue sky, white sand, ocean waves, casual beachwear, bikini top, shorts, standing pose, looking at camera, natural lighting, sunny day, relaxed atmosphere, beach background, palm trees, wide shot composition

<figure><img src="https://3286776735-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2F9ACKOAU4nFWSXPeE2B4s%2Fuploads%2FEGt6yBHAZyW8Dzq8PxxL%2Fimage.png?alt=media&amp;token=600b6d89-c592-43de-89d1-a6ad8066dfa5" alt="" width="375"><figcaption></figcaption></figure>

#### Example 2: Evening Pool Scene

**Good Prompt**: modern outdoor pool, night time, dark green skinny bikini, standing pose by poolside, ambient blue lighting, elegant evening atmosphere, modern villa background, subtle city lights, cinematic composition

<figure><img src="https://3286776735-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2F9ACKOAU4nFWSXPeE2B4s%2Fuploads%2F3lVvi8DMOemuUSNCWqd4%2Fimage.png?alt=media&amp;token=046abfc5-cb44-464c-a39a-4e9f19d1d766" alt="" width="375"><figcaption></figcaption></figure>

#### Example 3: Indoor Casual

**Good Prompt**: indoor setting, cozy bedroom, white tank top, blue jeans, sitting pose on bed, soft natural lighting, warm afternoon, relaxed atmosphere, modern room background, plants, natural composition

<figure><img src="https://3286776735-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2F9ACKOAU4nFWSXPeE2B4s%2Fuploads%2FLFFwqS2sjqtEFlusvPeL%2Fimage.png?alt=media&amp;token=a8fb9304-cd6e-4260-80ea-75ae76157691" alt="" width="375"><figcaption></figcaption></figure>

### Group Image Generation

{% hint style="info" %}
Group image generation works with **every companion** — main companions and custom characters alike, no setup needed.
{% endhint %}

Generate images with **up to 3 companions** together in a single scene. Group photos cost **50 Moments per companion** at 1K (Fast) or 1.5K resolution (doubled for 2K). Companions' own reference photos are included; extra gallery references you pick cost **10 Moments each**.

<figure><img src="https://3286776735-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2F9ACKOAU4nFWSXPeE2B4s%2Fuploads%2Ff2qqDS8JLqs8oeFqcIT8%2Fimage_1771633332550.webp?alt=media&amp;token=b1aef9f4-a31c-4158-983d-ec643cf1729e" alt="" width="563"><figcaption></figcaption></figure>

#### How to Generate Group Photos

1. **Open the Generate page** and select **Images**
2. **Select Group** (next to **Solo**)
3. **Select your companions** using the companion selector — filter by source, type, or gender
4. **Write a scene description** placing the companions together
5. **Choose your aspect ratio and resolution** (same options as Advanced solo mode)
6. **Generate** your group image

#### Group Image Pricing

The base cost is **50 Moments per companion** at 1K (Fast) or 1.5K resolution, doubled for 2K. Each selected companion's own reference photo is included:

| Companions   | 1K / 1.5K Cost | 2K Cost     |
| ------------ | -------------- | ----------- |
| 2 companions | 100 Moments    | 200 Moments |
| 3 companions | 150 Moments    | 300 Moments |

Extra **picked reference photos** add **10 Moments each**.

{% hint style="info" %}
**Including your persona requires Ultimate.** All Ultimate subscribers can select a saved persona under **You in the photo**, alongside one or two companions. There is a limit of **three people total**. Including the persona adds **10 Moments**, including at 2K: one companion plus your persona costs **60 Moments** at 1K/1.5K or **110 Moments** at 2K. See [Persona Photos](/create/persona-photos.md#include-yourself-in-a-studio-group-photo).
{% endhint %}

{% hint style="info" %}
Some companion pairings work better than others — experiment to find your favorites!
{% endhint %}

#### Group Photos in Group Chats

In group chats, the **instant image button** generates a group photo using the conversation context. The AI selects which companions should appear; the base cost is **50 Moments per companion** in the generated image. The chat's active persona can also appear when the scene calls for it, adding **10 Moments**. See [Persona photos in chat](/create/persona-photos.md#include-yourself-in-chat-photos).

***

### Editing Images

#### How Image Editing Works

You can edit generated images to create variations:

1. **Generate Page**: View your images
2. **Edit Option**: Look for edit/pencil icon
3. **Enter Edit Prompt**: Describe what you want to change
4. **Generate**: New image is created with your edits
5. **SFW only**: Image editing is only available for SFW images

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

<figure><img src="https://3286776735-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2F9ACKOAU4nFWSXPeE2B4s%2Fuploads%2FwMxbCPQq7HMubXG9yPWy%2Fimage.png?alt=media&amp;token=34b24067-f575-47c3-a8d6-d9e1c1bbb257" alt="" width="75"><figcaption></figcaption></figure>

#### Edit Prompt Tips

* **Be Specific**: Clearly describe what to change
* **Focus on Changes**: Only mention what's different
* **Use Same Format**: Use comma-separated tags format
* **Example**: "same setting, change dress to red, evening lighting"

<figure><img src="https://3286776735-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2F9ACKOAU4nFWSXPeE2B4s%2Fuploads%2FJDqqTRGRL1Vn9bsflrB7%2Fimage.png?alt=media&amp;token=bcc5f9eb-037b-42f5-af80-95b649c9dde8" alt="" width="375"><figcaption></figcaption></figure>

<figure><img src="https://3286776735-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2F9ACKOAU4nFWSXPeE2B4s%2Fuploads%2FPD26S9ibxJpYmVJ9HBQq%2Fimage.png?alt=media&amp;token=c4822a80-b0d7-4d9f-b7ab-6410af087ace" alt="" width="375"><figcaption></figcaption></figure>

### Image Refunds

#### Reporting a Bad Image

If an image generates successfully but the result is clearly broken or glitched, you can report it for review.

1. **Open the Image:** View the image you want to report.
2. **Click Report:** Select the report option on the image.
3. **Describe the Issue:** Briefly explain what went wrong.
4. **Submit:** Your report will be reviewed by our team.

If the report is reviewed and confirmed as a valid issue, you will be **refunded the moments** used for that generation.

{% hint style="warning" %}
**Policy update (July 13, 2026):** Advanced image generations and image edits can still be reported, but they are **no longer eligible for refunds**. This decision was made to ensure the sustainability of the platform given the increased costs of the new advanced image generation model. Reports on advanced images and edits are still reviewed and help us improve quality.
{% endhint %}

### Troubleshooting Image Generation

#### Image Not Generating

**Possible Causes**:

* Insufficient moments
* Service temporarily unavailable (check [Discord Integration](/extra/discord-integration.md))

#### Safety Check Messages

Every image request passes through our safety checks before and after generation. If a check stops your request, the **Generation stopped** or **Image Generation Failed** notice tells you exactly what was flagged and what to change. The most common messages:

* **Your prompt was stopped before generation.** The prompt itself was flagged, for example for possible underage content, a real person's likeness, hate speech, self-harm, graphic violence, non-consensual activity, or instructions to bypass the checks. Rewrite the flagged part and try again. Nothing is charged when a prompt is stopped before generation.
* **We couldn't create an image description that passed the safety check.** This applies to images requested in chat and instant images, where your companion writes the image description from the conversation. The notice names what was flagged, so steer the conversation toward a different scene.
* **The generated image was blocked by our safety check** (or **copyright check**). The image model rejected the finished image. Try a different scene, or a different character design if the copyright check was the one that stopped it. Moments for a generation that fails after it starts are refunded automatically.
* **A reference image was rejected by our safety, copyright, or privacy check.** One of your reference photos was flagged. The privacy check flags references that appear to show a real person. Swap in a different reference and try again.
* **The safety checker is temporarily unavailable.** This is an outage on our side, not a problem with your prompt. Wait a moment and try again; nothing is charged.

These notices describe what the check detected and how to adjust your request. They never show raw detector or provider details.

### Frequently Asked Questions

**Q: How much do images cost?**

* 50 moments per image in Premium mode and in Advanced mode at 1K (Fast) or 1.5K; 100 moments for 2K. Group images cost 50 moments per companion at 1K or 1.5K (doubled for 2K). In Advanced solo generation, reference photos after the first cost 10 Moments each. In groups, companions' own references are included; each extra gallery reference you pick costs 10 Moments. Ultimate subscribers can include a persona for 10 additional Moments per group photo. See [Persona photo costs](/create/persona-photos.md#moments-and-costs).

**Q: Can I use reference photos?**

* Yes! In Advanced mode you can add up to **5 reference photos** from your gallery to steer the generation. Your first reference is included in the base price; each additional one costs 10 moments.

**Q: Can Advanced mode generate NSFW images?**

* Yes — Advanced mode is now very capable of NSFW and intimate scenes, alongside its class-leading realism and likeness accuracy.

**Q: Can I generate in 4K?**

* Not yet — 4K is coming soon to Advanced mode. Current options are 1K (Fast) and 1.5K at the standard price, and 2K at double the price.

**Q: What's the difference between 1K (Fast) and 1.5K?**

* Both cost 50 moments per companion. **1K (Fast)** uses the model's smallest output size, so it renders the quickest, which is ideal for iterating on a prompt. **1.5K** gives you more pixels for the same price, and **2K** is the sharpest option at double the price.

**Q: My generation was stopped by a safety check. Was I charged?**

* No. Prompts stopped before generation are never charged, and a generation that fails after it starts is refunded automatically. The notice tells you what was flagged so you can adjust your prompt or references. See [Safety Check Messages](#safety-check-messages).

**Q: Can I use Advanced mode with any companion?**

* Yes — every companion supports Advanced mode with nothing to enable. Custom characters use their avatar as the reference automatically, or an optional [reference photo](/create/custom-characters.md#advanced-image-reference-optional) their owner has set in the Character Studio.

**Q: Can I request specific images?**

* Yes! Be specific about what you want in your request.

**Q: How do I get better images?**

* Be specific and include details about setting/clothing/pose.

**Q: Can I edit images?**

* Yes! Click on an image and use the edit feature to create variations.

**Q: Why did I get a different image than expected?**

* AI generation creates unique images each time. Be more specific or use edit feature.

**Q: Can I generate multiple images at once?**

* Yes. Premium and Advanced solo companion batches support up to **8 images**. On Ultimate, solo persona photos support **1, 2, or 4**. See [Batch Image Generation](#batch-image-generation) and [Persona Photos](/create/persona-photos.md#moments-and-costs).

**Q: How many images can I generate?**

* Unlimited, as long as you have moments.

**Q: Do images match my conversation?**

* Yes! Images are generated based on conversation context and your requests.

**Q: Can I download images?**

* Yes, you can download generated images from the images tab.

***

Ready to create videos? Check out [Video Generation](/create/video-generation.md) or learn about [Custom Characters](/create/custom-characters.md)!