Portrait editing
A wavespeed face swap starts with two distinct roles: a target portrait to keep and a source face to introduce. This guide helps you prepare both, write a precise edit request, and check whether the tool you open supports that workflow.
A face-specific request identifies who should change and what should remain. A general image request may leave both decisions open.
A client-approved composite needs a different expression while retaining the framing of the chosen shot.
Treat the chosen shot as the target and the approved expression as the source; check identity and lighting before using the result. If the real problem is softness, enlargement is a separate task.
wavespeed upscaleA still portrait is being prepared for a motion project.
Resolve the still-image edit first. A single edited frame does not establish that the same face will remain consistent throughout a clip.
wavespeed video converterA team is deciding whether face replacement or a broader image edit fits its approved concept.
Specify identity preservation for the first task; compare available model capabilities before assuming every image model accepts two portraits.
wavespeed ai modelsAn authorized restoration uses a clearer source portrait to repair a damaged face in an older photograph.
Keep the original photograph and document the edit. If the face is simply too small to inspect, assess resolution before attempting replacement.
wavespeed upscaleThese are the three decisions that make a face-replacement brief more specific than a general image prompt.
Identify the target as the image whose composition should remain, and the source as the person whose face may be used. Naming both roles prevents an ambiguous request to blend two portraits.
State whether the target's pose, hair, clothing, background, and lighting should stay intact. Decide which changes are acceptable before you judge an output.
Confirm permission to use both portraits. Review the result for recognizable identity, visual seams, and any misleading implication that the depicted event really happened.
Start with clear, permitted portraits, then check the destination for a suitable image-editing workflow. These adjacent guides help separate face replacement from other tasks.
Use this table to frame your request; it does not imply that every destination model offers a dedicated swap control.
Primary aim
Face-specific request
Change the face within a chosen target image.
General image request
Create or revise an image without a fixed identity-transfer goal.
Inputs to prepare
Face-specific request
A target portrait and a permitted source portrait.
General image request
A prompt and any reference images the chosen tool accepts.
Identity instruction
Face-specific request
Say whose face is the source and whose scene is the target.
General image request
Describe a subject or style without requiring a specific source identity.
Composition
Face-specific request
Ask to preserve the target's framing, pose, and setting.
General image request
Leave composition open or describe a new one.
Quality check
Face-specific request
Inspect facial alignment, skin transitions, and lighting.
General image request
Inspect whether the overall image matches the brief.
Before sharing
Face-specific request
Verify consent and disclose a consequential identity edit.
General image request
Review image rights and whether the result could mislead viewers.
A precise prompt cannot supply missing permissions, repair every source mismatch, or guarantee a particular tool's features.
The prompt describes the edit, but identity transfer depends on image inputs and on whether the selected tool accepts them.
Workaround
Have both permitted images ready, then verify the destination's input options before proceeding.
Tiny, blurred, or heavily obscured faces leave little information for a convincing match.
Workaround
Choose clearer source material and assess image quality as a separate step.
An acceptable portrait does not guarantee stable identity, expression, or lighting across moving frames.
Workaround
Evaluate a workflow intended for video rather than assuming a still-image result will transfer.
A convincing replacement can misrepresent a real person, especially in news-like, intimate, or commercial contexts.
Workaround
Get consent, avoid deceptive uses, and label material edits where viewers could otherwise be misled.
Open the destination to inspect its current image-editing options. Identify the target and source portraits, state what must stay unchanged, and confirm that the available workflow supports the inputs you need before expecting a face-replacement result.
Explore image editingPrepare a target portrait and a source portrait that you have permission to use. Describe which face should change and which parts of the target should remain, then check whether the tool you open supports the required image inputs.
Text can explain the desired edit, but it cannot provide a specific person's face on its own. For identity-based replacement, expect to need a usable source image and a target image if the selected workflow supports them.
Differences in head angle, lighting, expression, or image clarity can make the face look pasted in. Try better-matched, clearer portraits and inspect the edges, shadows, and skin transitions rather than judging only the thumbnail.
A still-image edit addresses one frame, not consistency across a sequence. Video requires a workflow that can account for movement and changing expressions; check the available tool's capabilities before starting.
Permission and context matter. Obtain consent for the images you use, avoid impersonation or deceptive presentation, and disclose significant edits when viewers could mistake the result for a real event.