How to face swap a video in Magic Hour: step by step

What you need

• A short video in which the target face is clearly visible.
• A clear replacement portrait, preferably at a compatible angle.
• Permission to use both the likeness and the footage in your intended context.
• A browser and a compatible upload; no local AI installation is required.

The guest workflow is separate from account credits and rewards. An old allowance expressed as “400 free frames” is not a current universal limit.

Step 1: Open the video tool

Go directly to video face swap. Make sure you selected video rather than photo or GIF face swap. Review the current input and output limits before choosing a long source clip.

Choose the free tool or the full editor

For a short trial, stay on the free Video Face Swap product page. Upload the video file and replacement portrait. This route does not use account credits; signing in does not give the same free tool a second daily allowance.

For a credit-based project, open the Face Swap Video editor and sign in. Check the account’s generation estimate and available export settings before submitting. The current route guide separates these experiences; a paid account does not make every setting on the simplified guest page available.

Magic Hour input workflow, captured September 25, 2026. This shows the upload or prompt controls, not a completed generation. View official source

Step 2: Upload a focused clip

Choose a sequence where the face stays visible. Motion blur, sudden cuts, strong profile angles, and hands crossing the face make the job harder. For the free guest workflow, keep the clip within ten seconds.

The route guide lists MP4, M4V, MOV and WEBM video inputs. Keep an original copy; if an otherwise valid upload fails, re-export a short MP4 sample rather than repeatedly uploading the full project.

Use the upload methods shown in the current interface. Do not assume that an older dashboard’s YouTube-import controls exist in the guest form.

Step 3: Add the replacement portrait

Choose even lighting and a face angle close to the source footage. Avoid filters that heavily alter facial geometry. Check any face-mapping controls for scenes containing more than one person.

Select the person before generating

A group scene needs an explicit mapping. On the free product page, switch from Single to Multi, wait for face detection, and assign a replacement portrait to each intended face. The free Multi experience supports up to three replacement faces; for more, use the signed-in editor.

In the full editor, check Apply to: Everyone uses the same replacement image for the faces, while Specific faces lets you assign replacements individually. Review the detected faces and each assignment before generating. For a group scene, uploading a video file is the supported route documented in the step-by-step guide. If only one person should change, do not leave an all-face setting selected.

Multi mode fixes the start position rather than exposing the same trimming controls as Single. Prepare the intended segment before uploading, then verify its opening and ending frames. This prevents a correct replacement from being applied to the wrong part of a longer video.

Step 4: Generate the swap

Confirm the inputs and any displayed charge, then generate. Keep the job open or use the status available in your account. Queue time and processing vary; a fixed “under three minutes” claim is not a reliable guarantee for every clip.

Step 5: Review and download

Watch the whole result, not just the first frame. Look for unstable eyes, distorted teeth, a shifting jawline, and blending errors during movement. If needed, change the portrait or source segment and compare the revision with the original.

For a no-signup generation, download from the result area on that product page before leaving. A guest result is not proof of a saved project in My Library. The route guide explains where that result appears; signed-in projects have a separate saved-project workflow.

Which plan fits a finished project?

Check current pricing for the selected tool, model, and output settings. Do not choose a tier solely from its label; finish a representative project and estimate revisions first.

Paid subscriptions and credit-pack purchases provide watermark-free output. Commercial use is permitted with those purchases or separately agreed paid usage-based billing under the Terms, including client work, subject to the rights in your source material. Credit packs do not unlock every subscriber-only setting. Choose paid access for the export requirements you actually need, rather than assuming a subscription is the only way to remove the watermark.

Improve a difficult result

Use a clearer source, a better-matched face angle, or a shorter unobstructed sequence. Review motion across frames. Upscaling the finished result cannot reliably repair a swap that used the wrong facial geometry.

When to use Smart or Stable face handling

Where the full editor offers face-handling controls, start with Smart for turns or temporary obstructions. Stable may better retain facial hair and face shape, but edges can flicker in some clips. These are tradeoffs in the troubleshooting guide, not guarantees that every accessory or facial detail will transfer. Compare the same short section after changing one setting.

If the result is blank, unchanged, or technically broken after checking the inputs and mappings, keep the project and send support the diagnostic details instead of repeatedly submitting the same request. Include the route, project link or visible ID, approximate time, mode, and where the problem occurs. A completed result you dislike is different from a technically failed generation.

Combine it with other workflows

For several people who need different replacements, use Multiple Face Swap Video and map each person separately. Use photo face swap for still images, lip sync to match speech in a video, or Talking Photo for a speaking portrait. Each tool has its own inputs and limits.

Face swap first, then lip sync

If the identity and speech both need to change, complete the face swap and download it first. Then open the Lip Sync editor and upload that downloaded video with the new audio. The current completion actions do not automatically pass the face-swap result into Lip Sync; choosing a YouTube source or preset there selects different footage.

If you need a different body, clothing or wider character appearance in the video, use Character Replace instead. Face Swap Video changes facial identity; it does not guarantee whole-head or hairstyle replacement. The face-swap guide distinguishes these jobs.

Publish with accurate context

Secure permission for the face and footage, check commercial terms, and apply any platform-specific disclosure requirements. A generated spokesperson should not be presented as a real customer giving a testimonial about an experience they never had.

Choose the next face-swap guide

Use the Magic Hour Face Swap overview to compare photo, video, GIF and multi-face routes. If a result flickers, blurs or blends poorly, follow the face-swap troubleshooting guide. To compare no-signup limits and watermarks across providers, use the free face-swap tools comparison.