Try-On
Transfer clothing and other wearable fashion items from one person image to another. Try-On preserves product details, texture, and fit while matching the target model’s pose and shape.
What is Try-On?
Virtual try-on shows how a specific item will look on a specific person's photo. Unlike Product to Model, which generates an on-the-spot AI fashion model to showcase the product, Try-On keeps your chosen model, pose, and body shape while swapping in the garment or accessory.
When to Use
Use Try-On when you want to swap clothing on an existing model photo while keeping that person’s identity, pose, and background. It’s ideal for connecting familiar models with new products—tops, bottoms, dresses, outerwear, shoes, hats, scarves, or other wearable fashion items. After you get a result, send it to Edit if you want to change pose or background for more variety.
- Starting from a standalone product photo without a person? Use Product to Model.
- Want the product on a specific person but do not need a particular photo or pose? Use Face Reference.
- Already have the product on a person and only want to change who it is? Use Model Swap.
How to Use It
Choose or Upload a Model Image
Upload a clear photo of the person who should wear the item. E-commerce style photos work best (studio background, even lighting, 2:3 or 3:4 aspect ratio).
Upload the Product Image
Use a flat-lay, ghost mannequin, or on-model photo (recommended). Works for garments and other wearables like hats, scarves, shoes, and bags. Show the full item with minimal shadows. Clear outlines and high resolution improve results.
Adjust Settings (Optional)
The top toolbar contains all generation controls. Most users can proceed with default settings, but you can fine-tune results using these options:
- Prompt (Optional): Add instructions to guide the generation, such as "open jacket", "tucked shirt", or other styling details.
- Number of Images (1–4): Generate multiple variations in a single run and select your preferred result.
- Category (Auto / Top / Bottom / Full-body): If the product image contains multiple items (for example, a top and bottom) and you want to transfer only one, select the category manually. When you select a category, the system automatically adds guidance to focus on that item type. Otherwise, leave it on Auto to let the system detect it.
Run the Try-On
- Click Run to start. The system will process your inputs, apply the product to the model image, and produce final visuals.
- Wait: Generation takes approximately 30 seconds for 4K resolution output.
- Review the Results: Browse through multiple results using the navigation arrows or thumbnails on the side.
Prepare for Export or Continue Editing
- Download, upscale, or use Export for Web in Gallery
- Fix mistakes, adjust background, or change aspect ratio with Edit and Reframe
- Create a short video from this image with Image to Video
Parameters
Only the model image and product image are required. Other inputs are optional.
| Parameter | Description |
|---|---|
| Model image | The person who will wear the item. Use a clear, front-facing or three-quarter pose with even lighting and a simple background. |
| Product image | The wearable item to apply. Flat-lay, ghost mannequin, or on-model photo with clear outline, texture, and print. On-model product photos work best (shows the item on a person). |
| Prompt | (Optional) Add instructions to guide the generation. For example: "open jacket", "tucked shirt", "rolled up sleeves". |
| Category | (Auto / Top / Bottom / Full-body) If the product image contains multiple items (for example, a top and bottom) and you want to transfer only one, select the category manually. Otherwise, the system will auto-detect it. |
| Number of Images | (1–4) Generate multiple variations in a single run and select your preferred result. |
Best Practices for Model Image
✅ Recommended
- Use a 2:3 aspect ratio for balanced framing and optimal model proportions.
- Ensure the outfit is fully visible, without being cropped or cut off.
- Choose fitted clothing so the system can clearly detect edges and body shape.
- Use three-quarter or full-body shots for best product alignment.
- Keep the pose natural and straightforward, such as standing or slightly angled.
❌ Avoid
- Extreme poses or unusual perspectives.
- Photos with multiple people or children.
- Hair or objects covering the clothes.
- Wide aspect ratios or distant shots.
Best Practices for Product Images
Follow these guidelines to achieve the most accurate product transfer and realistic fitting:
✅ Recommended
- On-Model photos deliver the best results, as garments and accessories appear in natural shape and lighting.
- Ghost mannequin shots work well when model photos are unavailable, providing clear structure.
⚠️ Acceptable but less optimal
- Flat-lay images can be used, but depth and fit information are limited, which may reduce realism, especially for structured items like shoes or bags.
❌ Avoid
- Apparel on hangers: these introduce perspective distortion and lose body shape context.
- Apparel on the floor: these appear wrinkled or uneven, making fitting and texture mapping unreliable.
Using clear, structured product photos ensures cleaner segmentation, better draping accuracy, and more consistent visual quality.
Troubleshooting
- Original clothing is not being removed correctly → Add a specific instruction to remove the unwanted item (e.g. "remove the jacket").
- Product details are not transferring accurately → Use a higher quality source image with clear outlines and good lighting. On-model photos typically produce the best results.
- Results don't match expected styling → Add instructions in the prompt field to guide the generation (e.g., "tucked shirt", "open jacket").
Where Next?
- Improve the resolution of your image with Upscale in Gallery
- Apply Reframe to fit marketplace or social aspect ratios
- Convert your new image into motion clips with Image to Video