Photo workflow
How to face swap a photo
Follow the five-step target-and-reference workflow, then check pose, lighting, hair, occlusion, and edges.
Open the step-by-step guidePractical field guides
Make better face swaps before you spend credits
These guides turn the variables that most affect a face swap into repeatable checks. Choose a workflow, prepare the media, and diagnose the result without guessing.
Choose your workflow
Start with the constraint you need to solve
Each guide is written around a real production decision: selecting inputs, preparing motion, keeping a photo set consistent, or deciding whether media is safe to upload and share.
Group photos
Multiple face swap
Detect, select, and map several identities in one group photo without losing track of who replaces whom.
Open the group-photo guideMotion
Video face swap
Prepare clips for turns, movement, blur, occlusion, continuity, and a predictable credit estimate.
Open the video guidePhoto sets
Batch face swap
Keep identity, framing, light, and quality-control decisions consistent across multiple images.
Open the batch guideShort loops
GIF face swap
Scout loop seams, fast turns, blur, occlusion, and frame-to-frame identity changes before processing.
Open the GIF guideResponsible use
Privacy and consent
Check permission, retention, training use, account controls, and sharing risks before uploading a face.
Open the privacy guideImage exports
Trial watermark and paid unlock
See exactly when an image displays the complete account email, what a purchase changes, and why an old file stays unchanged.
Open the watermark guideVerified rewards
Free credits, check-ins, shares, and referrals
See the exact +1-credit rules, UTC resets, referral cap, duplicate protection, and trial image export boundary.
Open the free-credit guideChoose a workflow
Batch vs multiple face swap
Choose between multiple output images and mapped replacements inside one group photo using current limits and costs.
Compare the workflowsPrivate preflight
Image input readiness checker
Measure dimensions, luminance, contrast, clipping, and edge detail locally without uploading the selected image.
Open the local checkerOriginal research
Controlled input benchmark
Inspect six downloadable variants, exact transformations, formulas, measurements, hashes, and evidence limits.
Review the benchmarkOutput evaluation
Face swap quality scorecard
Rate identity, blend, pose, lighting, occlusion, technical integrity, and motion stability with one transparent rubric.
Open the scorecardWorking principles
Four checks that prevent most avoidable failures
Use a clear identity reference
Choose a sharp, unobstructed face with enough detail around the eyes, nose, mouth, jaw, and skin texture.
Match pose and expression
A reference that resembles the target angle and expression gives the model fewer conflicting cues to resolve.
Inspect the difficult frames
For video and GIF work, review fast turns, hands over the face, profile views, blur, and lighting transitions.
Test before scaling
Validate one representative image or a short clip before committing a full batch or longer video.
Evidence and limits
Useful guidance without invented benchmarks
The Product Team derives these checks from the current DeepSwapAI workflows and the observable media variables users can control. Quality still varies by input, and no guide can guarantee a specific output.
Current product behavior
Workflow costs, upload limits, retention, and watermark rules are checked against the current product and documented in the verification methodology.
Direct product paths
Use the photo workspace, batch workspace, group-photo workspace, video workspace, or GIF workspace when you are ready to test.
Prepare one representative input
Start with the face swap quality guide, then choose the exact workflow and review its cost, privacy, and image-export rules.