Create new scenes from one familiar face
Use one authorized reference photo to create an AI portrait in a new scene, guided by visible features. This is a newly generated image, not a pixel-perfect background replacement or a record of real events.
No app required · Mobile and desktop browsers
Use a single-person photo with the necessary usage rights and authorization.
Start with an everyday setting, or describe clothing and poses.
One delivered image uses one credit. Failed tasks release the credit.
Accounts, Credits, and Privacy
Registration and Credits
Each new regular account receives one gift credit. Registration accepts only the mainstream personal email domains listed on the signup page. This deployment does not require email verification for new accounts; existing unverified accounts must still verify.
Submitting a task holds one credit; successful delivery consumes it and failure releases it. Uploads, previews and downloads incur no extra credits. Additional credits may be issued after administrator verification. There is no online checkout, subscription or auto-renewal. Unused credits from verified payments may qualify for a manual refund under published rules; gifts and compensation are not redeemable for cash.
Requested size 1024×1536；actual provider dimensions are delivered and may differ, without stretching or cropping. The result page shows the actual dimensions. Supported output formats: png jpeg webp
Data Processing and Retention
Unsubmitted uploads are cleaned after 1 hours. References are retained for 1 days from upload (active tasks keep private working bytes); results for 30 days from saving; descriptions for 30 days.
Private photos are not public by default, and this website does not proactively use them for training or promotion. See the privacy policy for provider data purposes and processing regions.
Read the full privacy and provider disclosuresWhat makes a good reference?
For better feature retention, we recommend using a recent, single-person photo with even facial lighting, clear features, and no severe occlusion.
Unshown body parts (like height and physique) will be inferred by the model during generation and cannot be considered accurate restoration. If physique is important, choose a photo that shows those areas.
An exact likeness is not guaranteed. New angles and poses affect results; unseen body parts are inferred, not a record of actual events. We do not have independently approved public cases yet, so we do not display a gallery of real generated cases.