Image Generation — Reference Images & Prompt Tips
Summary: This guide explains how to upload a reference image to reflect its composition and color palette, how to use the Prompt Enhancement and Image To Prompt features, and how automatic face enhancement (ADetailer) improves portrait quality.
Using Reference Images
- Click + Edit from image in the options row to upload an image or sketch as a reference — you can also drag and drop it, or paste it from your clipboard. Uploading a reference image switches the interface to the Edit tab.
The Edit tab, where your image goes into Source Image · required. Add a Reference sketch if you like, describe the change ("Describe what you want to change."), and click Apply.
- Work on your image in the Edit tab. For details, see Editor.
Image Detail View
Click a generated image to open the detail view. The right-hand panel shows the model, resolution, creation date, and prompt, plus Use, Send, and Organize buttons.
Image Action Buttons
Result card and action buttons.
- Hovering over an image shows Copy / Download / Star at the top right, ↳ (Add as reference) at the bottom left, and the image ⋮ menu at the bottom right.
- The bar below the card shows the label (Generate/Edit), number, author avatar, ▾ (expand to see the engine, prompt, and reference images), Like / Dislike, Comment, time, and the card ⋮ menu.
- Image ⋮ menu: Use › (Upscale / Remove BG / Animate / Edit in Canvas / Filter) | Add as reference / Download / Share | Star / Save to Library | Delete
- Card ⋮ menu: Copy Settings / API Code | Move To / Duplicate / Compress / Select multiple | Delete Contents
For details on each menu and the detail view, see Result Card ⋮ Menu and Image Enhancement.
Prompt Assistant (Robot Icon)
Click the robot icon at the bottom of the Direct panel, to the left of the run button, to open the prompt assistant menu. It has two items, Prompt Enhancement and Image To Prompt, each costing 1 credit.
Robot icon menu — Prompt Enhancement / Image To Prompt. The robot icon sits next to Run in the bottom row of the Direct panel.
Prompt Enhancement
- Use this when you want to make your prompt more specific and detailed.
- The enhanced prompt is filled into the input box.
Image To Prompt
- Upload an image to extract a prompt that describes it, then use that prompt for generation.
Prompt extracted from an image (example).
Face Auto-Enhancement (ADetailer)
XBRUSH automatically applies ADetailer during image generation to detect and refine faces in generated images. This process runs in the background without any manual configuration.
- What it does: Detects faces in the generated output and applies an additional enhancement pass to improve facial detail, proportions, sharpness, and consistency.
- When it applies: ADetailer activates automatically whenever a face is detected in the generated image.
- No setup required: There are no settings to configure — ADetailer runs silently and the result is the enhanced image you see in the results feed.
This is especially beneficial when generating portraits, group scenes, or any image where facial accuracy matters. Without ADetailer, faces in AI-generated images can appear blurry or distorted; ADetailer corrects this automatically.
Generation History
See "Credit Usage History."
Next Steps
- Enhancement — Upscale for higher resolution, inpaint for detailed edits
- Video Generation — Convert a selected image into a video
- Social Media — Publish results to your social accounts