Sign up free: 10 credits every day

How to Make a Face Swap Video: Step-by-Step Guide

November 15, 2026
Learn how to make a face swap video in 5 steps: pick the right source photo and clip, fix flicker, keep the audio and export clean results with AI.
How to Make a Face Swap Video: Step-by-Step Guide
guides
video face swap
face swap

A face swap video is one of the most shareable things you can make with AI: your face in a movie scene, a friend dropped into a music video, or yourself in a travel clip you missed. Five years ago this took a desktop deepfake program, a strong graphics card and days of training. Today you can make a face swap video in a browser in a few minutes. This guide shows how to make a face swap video step by step with the aideepfake.io video face swap, how to choose footage that swaps cleanly, which settings fix flicker, and how to export a result that looks as good as the clip you started with.

What you need to make a face swap video

Every face swap video is built from two inputs:

  • A source face: a still photo of the person whose face you want to use. It sets the identity for every frame of the result.
  • A target video: the clip that keeps its body, background, motion and audio. Only the face changes.

On aideepfake.io the source must be an image (JPEG, PNG or WebP, under 10MB) and the target video must be an MP4 or MOV file up to 100MB. Clips can run up to 300 seconds, or up to 90 seconds when face enhancement is switched on. You do not need to install anything, and you do not need a powerful computer, because the processing happens on the server.

You also need the right to use both. Use footage you shot or have permission to edit, and a face that belongs to you or to someone who agreed to it. More on that at the end.

How to make a face swap video in 5 steps

Step 1: Pick a strong source photo

The source photo matters more than any setting. Choose a sharp, well-lit, front-facing photo of one person with both eyes visible and nothing covering the face: no hands, hair across the eyes or sunglasses. A neutral or slightly smiling expression works best, because the swap has to hold up while the face in the video talks, blinks and turns. A blurry or heavily filtered selfie will produce a blurry face in every single frame.

Step 2: Trim the target clip before uploading

A face swap video is priced and processed per clip, and a longer clip costs more credits and takes longer. Cut the video down to the shot that actually carries the idea before you upload it. A ten-second clip usually comes back in about a minute; a two-minute scene takes far longer. The tool also has trim frame start and trim frame end controls that narrow processing to a range of frames, but the credit cost follows the file you upload, so trimming the file itself is what saves credits.

Step 3: Upload both files

Open the video face swap tool. Upload the source photo into the source face box and the trimmed clip into the target box. If you just want to test the look on a single frame first, take a screenshot from the clip and run it through the photo face swap. A still swap costs 15 credits and finishes in seconds, which is a cheap way to confirm the source photo works before spending credits on video.

Step 4: Choose your options

For most clips the defaults are fine. Two optional switches are worth knowing about:

  • Face enhance rebuilds fine skin texture inside the swapped face. It is useful when the source photo is slightly soft.
  • Frame enhance upscales the whole clip.

Neither enhancer costs extra credits, but both add processing time to every frame, and that multiplies across the length of the video. Switch them on deliberately rather than by habit. The advanced panel holds the face detection, face selection and output encoding settings covered below.

Step 5: Run the swap and download

Click Swap Faces. A video swap costs from 20 credits, and most clips are ready in one to five minutes. The page checks for the result automatically. When it is done, download the file at full resolution. The original audio track is kept, so speech and music stay in sync with no extra editing.

How AI video face swap works

It helps to know what happens behind the button. Each frame of the clip is decoded, scanned for faces, mapped with facial landmarks, swapped with the source identity, blended back into the frame and written into a new video file. On aideepfake.io every frame gets its own detection pass rather than a track carried forward from the first frame, which is why hard cuts and brief occlusions do not derail the whole clip. The trade-off is that detection quality decides stability: if a single frame fails detection, that frame passes through unswapped and the result appears to flicker. If you want a deeper look at the technology, read how deepfake technology works.

Choosing footage that swaps cleanly

A disappointing face swap video is nearly always an input problem. The model is the same on every job; what changes is the footage. Before you spend credits, check the clip against these points:

  • Face size: the face should stay reasonably large for the whole trimmed range. A face that fills the frame at the start and shrinks to a dot in a wide shot will swap well for half the clip and break down for the rest.
  • Head angle: front-facing and three-quarter views swap best. Long stretches of full profile are the hardest material.
  • Motion blur: fast whip pans and shaky handheld footage blur the face, which makes detection less reliable.
  • Lighting: even lighting gives a natural skin tone. Hard shadows across half the face are a common cause of missed frames.
  • Obstructions: microphones, hands, hats and hair crossing the face interrupt the swap for those frames.
  • Number of people: a single person is easiest. With several people in the shot you need to tell the tool which face to replace (see below).

Many of the same rules apply to still images, and our realistic face swap tips cover matching angle, light and expression in more detail.

Fixing flicker and other common problems

The face flickers or disappears on some frames. Flicker almost always comes from detection, not blending. On those frames the face scores just below the detector threshold, so they pass through unswapped. In the advanced panel, lower the detector score so borderline frames still register. If the face is small in the frame, raise the detector size as well.

The swap jumps to the wrong person. In a clip with several people, the face can be handed to someone else when your subject leaves the frame. Set the face selector to reference mode and choose a reference frame number where your subject is clearly visible, so the match stays locked to that person for the whole clip. If you want everyone in the shot to get the same new face instead, the tool can replace every detected face with the one source face. Giving each person a different face in a single pass is not supported.

The result looks softer than the original. Every face swap video is re-encoded. In the advanced panel, set the output video quality high and keep the encoder preset sensible, especially when the original footage was clean. If the source clip itself is soft, frame enhance can upscale it.

The motion looks jerky. Leave the output FPS alone unless you have a concrete reason to change it. Changing the frame rate forces a resample that can introduce judder, which reads as a swap problem even when the face work is fine.

The skin tone does not match. Pick a source photo with lighting closer to the target clip. Warm indoor light on the source and cold daylight in the video is the most common mismatch.

Make a face swap video with a celebrity face

Want a famous face rather than your own? The celebrity face swap library has dedicated pages where the source face is already loaded and locked, so you only upload the target clip. For example, the Keanu Reeves face swap page lets you drop your own clip into a long-coat, action-movie look. Video swaps on celebrity pages cost the same 20 credits per clip. For film-style recasts, the movie face swap page collects ideas for putting yourself into a famous scene. Always label celebrity edits as fan-made AI parodies when you share them, and never present them as real footage.

Free credits, watermark and pricing

Every account can claim 10 free credits a day through the daily check-in. Because a clip costs from 20 credits, free credits alone do not cover a video, so the free video face swap page explains how check-in credits combine with purchased credits. Results carry an aideepfake.io watermark by default. Any completed purchase, either a credit pack or a subscription, unlocks a switch that turns the watermark off for new jobs, as explained on the face swap no watermark page. Plans and credit packs are listed on the aideepfake.io pricing page, and purchased credits never expire.

Automating video face swaps with the API

If you are building an app or need to run swaps from your own workflow, the face swap API accepts the same photo and video jobs as the web tool, with the same options for enhancement and watermark. It is a good fit for products that let their users put their own face into template clips.

Use face swap videos responsibly

Moving footage is the most persuasive format for a deepfake, which is exactly why it needs the most care. Only use faces you have permission to use. Never create sexual content of a real person, never create anything involving minors, and never use a face swap video to deceive, defraud, harass or impersonate someone. Do not present a swapped clip as a genuine recording of something that never happened. Label your edits as AI edits when you post them.

FAQ

How long does it take to make a face swap video?

Most clips are ready in one to five minutes on aideepfake.io. Short, trimmed clips are faster; long clips with face or frame enhancement switched on take longer because every frame is processed.

Can I make a face swap video for free?

Every account gets 10 free credits a day, and a video swap starts at 20 credits, so a clip usually needs some purchased credits on top. You can test your source photo for less with a photo swap first.

What video formats are supported?

The target video can be MP4 or MOV, up to 100MB and up to 300 seconds long (90 seconds with face enhancement on). The source face must be a still image.

Does the face swap video keep the original sound?

Yes. The audio passes straight through, so speech and music stay in sync with the new face.

Can I swap faces in a video with several people?

Yes. You can lock the swap to one person with reference mode, or replace every detected face with the same source face. Different faces for different people in one pass are not supported.

Why does my face swap video flicker?

Usually some frames miss detection because of blur, profile angles or shadows. Lower the detector score in the advanced panel, and trim away the hardest stretch of footage if needed.

Start your first face swap video

Pick one sharp photo, trim one short clip and run it through the aideepfake.io face swap tool. Your first face swap video will take a few minutes, and once you have seen how footage and settings affect the result, every clip after that gets better.