1. Show one clearly visible person
Single-person photos reduce confusion about whose clothing should change. Do not use a photo if subject suitability is unclear or the surrounding context prevents a reliable safety assessment.
Group photos also create technical problems: a model may alter the wrong outfit, merge limbs, or change people who were not meant to be edited.
2. Keep the relevant body area in frame
Use a medium, three-quarter, or full-body photo where the existing outfit is visible. Very tight headshots do not give the model enough context, while a subject who is tiny in a wide landscape frame may lose facial detail. Avoid cutting through shoulders, waist, hips, or knees at awkward points.
3. Prefer soft, even lighting
Daylight near a window or open shade is ideal. Strong backlight can turn the body into a silhouette. Flash hotspots and deep shadows can be mistaken for skin, fabric, or body boundaries.
Good lighting does not need to look professional. It only needs to make the face, clothing edges, and pose easy to understand.
4. Start with a straightforward pose
A relaxed front-facing or slight three-quarter pose is usually easiest. Crossed arms, hands on the waist, sitting with overlapping legs, or objects held in front of the torso create occlusions. The model has to invent what it cannot see, which raises the chance of distorted hands, straps, or proportions.
5. Use a sharp, unfiltered image
Choose the original photo when possible. Messaging-app compression, screenshots, beauty filters, artificial blur, and heavy color grading remove details that help the model preserve identity. JPG, PNG, and WebP all work; visual clarity matters more than the format.
A practical target: the person's face should still look clear when the image fills a typical phone screen.
6. Keep the background simple
A plain wall, uncluttered room, or ordinary outdoor setting helps the editor separate the person from the scene. Mirrors, patterned fabric, crowds, railings, and strong reflections can produce duplicated limbs or altered surroundings. The background need not be blank—just readable.
7. Check clothing edges and obstructions
The editor needs to understand where the current outfit ends and where skin, hair, hands, and background begin. Jackets tied around the waist, bags crossing the torso, scarves, long hair over the shoulders, and hands placed over fabric make those boundaries less certain.
You do not need a studio pose. Before uploading, look at the thumbnail and ask whether a person could quickly trace the outfit outline. If the answer is no, choose another frame or crop out distractions without cutting through the relevant body area.
8. Choose a photo you are comfortable processing
Even a privacy-conscious AI workflow sends the image to technical providers. Do not upload identity documents, private medical images, exact home addresses, or other highly sensitive details. Crop unnecessary bystanders and personal information before uploading.
Our Privacy Policy explains the processing path, lifecycle requests, and limits. Use the tool only if that handling is acceptable to you.
A 30-second pre-upload checklist
- Is there one intended subject?
- Are the face and current outfit sharp enough to inspect?
- Are the hands, hair, and clothing edges readable?
- Is the light even enough to separate fabric from the background?
- Does the crop include every area the new outfit must cover?
- Have unrelated people and private details been removed?
If all six answers are yes, the image gives a generative editor a cleaner visual brief. If a result still looks wrong, work through the result-quality troubleshooting steps.
When a result is not right
Try a different source photo instead of repeatedly sending the same difficult one. A clearer pose or better light is usually more effective than another attempt. Discard any result that changes identity too much or produces unsafe, transparent, or implausible clothing.