Can I put my friend's face or a celebrity's face on my video?

Not with Stand-in, and you should not anywhere. Putting a real person's face on a video without consent is what deepfake laws target: the US TAKE IT DOWN Act (May 2025) and Denmark's proposed likeness rules are two examples. Every Stand-in is a generated person who belongs to no one.

How much does an AI face swap cost per 30 seconds?

Stand-in is from $3.99 per 30 seconds (Lite, with a $9.37 credit pack). Higgsfield's Seedance 2.5 route is about $10-$16 per 30 seconds at its October 2026 credit prices. Kling's Standard plan works out to about $3-$4 per 30 seconds of generated clips. HeyGen's face swap is on plans from $24 a month and takes clips up to 15 seconds.

Does the swap change my voice?

Only if you ask. Your recording's audio stays as it is by default, because the swap changes pixels, not sound. Stand-in lets you keep your voice or swap it in one tap at upload. Close friends may still recognize a kept voice; strangers will not.

A deepfake, in the way the law uses the word, puts a real person's likeness on something they did not do. A performance swap puts a fictional person onto something you did do, in your own footage, with a label on it. Same technology family, opposite purpose.

Why does my swap look wrong around the hands and mouth?

Hands crossing the face and fast mouth shapes are the two hardest things for any swap engine. Refilm with hands below the chin and talk at normal speed, or pick the Standard or Pro level. If a Stand-in video comes out wrong, it is redone free or the credits come back.

Do I need an AI label if only the face changed?

Yes. YouTube, TikTok and Instagram all ask for a label on realistic AI-altered video of people. The size of the edit is not the test; whether the result looks real is. YouTube exempts only minor edits like filters and color changes. Turn the label on when you post.

Can I swap a video that is already posted?

You can download your own video, swap it and repost it. Note that Instagram's AI label is set while the post is a draft, so the swapped version needs to be a new post, not an edit. On YouTube and TikTok you can also set the label on a fresh upload.

How long does a swap take?

Stand-in videos are usually ready in about 10 minutes: roughly 4 minutes of render per piece on Lite, 6 on Standard, 7 on Pro. Longer clips take longer. Prompt-based tools take a few minutes per 5-15 second clip plus the time to write and retry prompts.

Guides · Last updated November 7, 2026 · 14-minute read

How do I AI face swap my own videos (and keep my performance)?

Film a normal phone video, upload it, and let a fictional person take your place while your movements, timing and words stay. That is what Stand-in does: film yourself, a fictional person takes your place, your performance stays. It costs from $3.99 per 30 seconds (Lite, with a $9.37 credit pack), takes about 10 minutes, and only works on your own footage. You cannot upload a real person's face.

Swap yourself in your own video See every option

Key facts

• Face-swap apps put a face onto a clip; whole-person replacement changes face and body in your own footage and keeps everything else.
• Never swap a real person's face without consent. The US TAKE IT DOWN Act (May 2025) and Denmark's likeness bill show where the law is going.
• Fast movement, hands over the face, groups and singing close-ups are what break swaps; a steady, standing, talking clip works best.
• Stand-in costs from $3.99 per 30 seconds (Lite, with a $9.37 credit pack); Higgsfield's Seedance 2.5 route is about $10-$16 per 30 seconds; Kling's own plan is about $3-$4 per 30 seconds for generated clips.
• A swapped real performance needs the AI label on TikTok, YouTube and Instagram. The platforms say the label alone does not cut reach.
• HeyGen's face swap takes clips up to 15 seconds and starts at $24 a month; it is built for stock avatars, not your performance.

Best for

• Creators who film themselves and want their face off the internet.
• Talking videos: opinions, stories, reviews, UGC, teaching.
• Steady, standing or seated clips with one person in frame.

Not for

• Putting a real person's face on any video. Stand-in does not allow it.
• Dance, fast arm movement, group shots and singing close-ups.
• Making a digital twin of yourself: no twin is ever made.

What is the difference between a face-swap app and replacing yourself in your own video?

A face-swap app takes a face from a photo and pastes it onto a face in a clip. The clip is usually someone else's: a meme, a movie scene, a stock avatar. HeyGen's AI Face Swap, for example, asks you to upload one clear photo and a short clip up to 15 seconds, and maps the photo's features onto the clip's face. Its page pitches it as "no filming." The direction is: your face onto their video.

Replacing yourself in your own video is the reverse. You film the clip. Then a fictional person takes your place in it. Your room, your clothes, your lighting, your gestures, your pauses and your words stay. The direction is: a new person onto your performance.

The two tools solve different problems. A face-swap app is for fun or for putting your face on a stock presenter. Whole-person replacement is for people who perform well but cannot show their face. The rest of this page is about the second one, because that is the one almost nobody explains.

Is it legal to face swap a video? What about consent?

Swapping a face in your own footage with a face that belongs to nobody is the safe case. The dangerous case is putting a real person's face on a video without their consent. That is what the word deepfake means in law, and the law is moving fast. The US TAKE IT DOWN Act, signed in May 2025, is the first federal law aimed at harmful deepfakes; it covers non-consensual intimate images, including AI-made ones, and gives platforms 48 hours to remove reported material.

Denmark has proposed amending its Copyright Act so that realistic imitations of a person's face, voice or movements cannot be published without consent; a bill was introduced in October 2026 to apply from 2027. The EU AI Act's transparency duties for AI content apply from August 2, 2026.

The rule for creators is simple. Your own footage, your own performance, a face that is not a real person's. Stand-in follows this by design: every Stand-in is generated and belongs to no one, you cannot upload a real person's face, and no digital twin of you is made.

Face-only swap or whole-person replacement: which should you use?

A face-only swap keeps your body, hair, neck and skin and changes the face region. It is quick and cheap, and it fails in a predictable way: the face does not match the neck, the hairline shifts, and anyone who knows you can still recognize your build, your hands and your posture. For privacy, that is a weak disguise.

Whole-person replacement changes the face and the body together, so the person on screen has their own skin tone, hair, build and hands. Your movements drive that person. Your clothes can stay (the Stand-in wears what you wore) or you can pick from ten outfits. The result is designed to play like a normal phone video, not a filter.

Pick face-only when the goal is a joke or a quick effect. Pick whole-person when the goal is that nobody can tell it was you. Stand-in does whole-person replacement only, for from $3.99 per 30 seconds (Lite, with a $9.37 credit pack), with 16 ready-made Stand-ins free to use or one you create and keep on every video.

What breaks a face swap?

Four things, in order of how often they ruin a result. First, fast movement: dance, arm swings, jumping and whip pans. The engine cannot keep the new face and body locked when the source moves faster than it can track. Second, hands over the face: touching your chin, pushing hair back, holding a mic close to your mouth. The hand and the face fight over the same pixels.

Third, groups. Two or more people in frame means the engine must decide who to replace and keep the others untouched; results get worse with each extra person. Fourth, singing close-ups. Mouth shapes in singing are bigger and faster than in speech, and a tight frame magnifies every error.

Other things that hurt: very dark rooms, strong backlight, the face leaving the frame, and sunglasses or heavy headwear. What works best is boring on purpose: one person, standing or seated, good light on the face, talking at normal speed, hands below the chin most of the time. That describes most opinion, review and UGC videos.

How much does it cost to face swap a 30-second video?

Prices below were read from the tools' pages in November 2026 and are per 30 seconds of finished video. Stand-in is from $3.99 per 30 seconds (Lite, with a $9.37 credit pack). A $9.37 pack buys 1,000 credits, and each started 15 seconds uses credits. Plans start at $19.99 a month and unused credits roll over. A failed video returns its credits; one that comes out wrong is redone free.

Higgsfield sells access to several video models by credit. Its own pricing post (verified October 2026) lists Starter at $15 a month for 200 credits and Plus at $49 for 1,000, and prices a 30-second Seedance 2.5 clip at 720p at 210 credits. That is about $10 per 30 seconds on Plus and about $16 on Starter, before any retries.

Kling's Standard plan is $10 a month for 660 credits. Third-party guides put Kling Video 3.0 at 6 credits per second at 720p without audio and 9 with audio, so 30 seconds of generated video is about $3-$4, made in several short clips, since Kling clips top out at 10-15 seconds. Kling's own guide lists 4K at 30 credits per second.

Whole-video routes compared, per 30 seconds (prices read Nov 2026)
Tool	What it does	Cost per 30 s	Clip limit	Keeps your performance?
Stand-in	Fictional person replaces you in your own footage	from $3.99 per 30 seconds (Lite, with a $9.37 credit pack)	Up to 10 minutes	Yes
Higgsfield (Seedance 2.5)	Generate or re-render a scene from references	about $10-$16	30 s per take	Partly, depends on the mode
Kling (Video 3.0)	Generate a clip from text, an image or a motion reference	about $3-$4	10-15 s per clip	Partly, with a motion reference
HeyGen AI Face Swap	Your photo onto a stock avatar or short clip	from $24 a month	15 s per clip	No

What are the quality tiers, and which one do you need?

Stand-in has three levels. Lite is the lowest price and renders in about 4 minutes per piece; it is right for steady, well-lit talking clips. Standard is the default, about 6 minutes of render, and holds up better on head turns and hand gestures. Pro adds the sharpest output, about 7 minutes of render, for videos you will reuse as ads or pin to a profile. The price shown on the site, from $3.99 per 30 seconds (Lite, with a $9.37 credit pack), is the Lite price with a $9.37 credit pack; Standard and Pro cost more per 30 seconds, and the exact credits for each level are shown before you pay.

The app checks your clip before you pay. If the movement is too fast for Lite, Lite is greyed out with the reason, so you are not charged for a level that will fail on your footage.

A good rule: start on Standard for anything with gestures, drop to Lite for a static piece to camera, and use Pro when the video is the one people will see first. HD and captions are included on plans; on packs, captions are 50 credits per 30 seconds.

How do you face swap your own video, step by step?

The whole process is five steps and most of the time is waiting. Film, upload, pick a person, wait, post with the label. The steps below are for Stand-in; other whole-video tools follow the same shape but ask you to write prompts or supply reference images instead of just uploading.

Before you film, read the section above on what breaks a swap. The best single tip: set the phone down. A steady frame does more for the result than any setting. Film in one take if you can; cuts are fine, but each new angle is a new problem for the engine.

After the swap, watch the whole video once at full size, not a thumbnail. Look at the mouth on hard consonants, the hands when they cross the body, and the hairline when you turn. If something is off, ask for a redo; it is free or the credits come back.

Can you keep your own voice when you swap your face?

Yes. The voice in your recording is audio, and a whole-person swap changes pixels, so your voice passes through untouched unless you ask to change it. Stand-in gives you the choice at upload: keep the voice from the clip, or swap it in one tap for a different one. Many people keep their voice for the first videos and switch later once they decide how private they want to be.

If you keep your voice, remember that a voice is also a way to recognize someone. Close friends will know. Strangers will not. If your reason for swapping is a job rule or a wide audience, keeping the voice is usually fine. If the reason is a specific person you are hiding from, swap the voice too.

Platform note: YouTube's help page says cloning your own voice for a voiceover does not by itself need disclosure, but a realistic AI person on screen does. So the label depends on the face, not the voice.

Do you have to label a face-swapped video as AI?

Yes, on all three big platforms. YouTube requires disclosure of AI-generated or altered content that seems realistic, and gives the example of making a real person appear to say or do something they did not. You answer the AI use question under Attributes when you upload. The page says disclosing alone will not limit the audience or the video's eligibility to earn money, and that creators who repeatedly skip it can get a label applied for them and face penalties.

TikTok's Community Guidelines say it requires creators to label AI-generated or significantly edited content that shows realistic-looking scenes or people. There is an AI-generated toggle on the post screen. Instagram's help page says labeling is sometimes required and that Meta adds an AI info label when it detects AI signals or when you disclose. Since August 31, 2026, a profile that features an AI-generated person also needs Instagram's AI-generated profile label or it can lose recommendations.

A swapped performance shows a person who does not exist. Label it. Stand-in never helps hide that a video is AI.

Will people be able to tell?

Sometimes, and that is fine, because you are labeling it anyway. The goal of a performance swap is not to fool anyone. It is to let a real delivery reach people without your face attached. Viewers already assume a lot of video is AI: 83% of consumers in Animoto's January 2026 survey said they had watched a video they suspected was AI-made.

What viewers punish is stiffness, not disclosure. A 2,000-person 2026 study found avatar speakers rated less trustworthy than unaltered video (0.42 against 0.50 on its scale). A swapped clip is not an avatar: the timing, the eye line and the gestures are a real person's, so the thing that study measured is still there. We do not have platform data to prove a swapped clip holds attention like a filmed one; the studies suggest the delivery is what matters.

Where swaps give themselves away is on the failure list: fast hands, mouth close-ups, groups. Keep the footage steady and the swap stays in the background of the viewer's attention.

Can you keep the same fictional person across all your videos?

Yes, and you should if you are building a channel. Stand-in lets you create your own Stand-in once and use that same face on every video, so your audience sees one person over time. The 16 ready-made Stand-ins are free to use and can also be reused on every upload. Either way, the person belongs to no one real.

Consistency matters more than people expect. A channel where the face changes every week reads as random clips; a channel with one face reads as a creator. If you make UGC for brands, the opposite can be useful: a different Stand-in per brand keeps the videos from pointing back to each other.

Clothes follow the same logic. Ten outfits are available, or the Stand-in wears what you wore. Pick one look for your own channel and let it vary for client work.

What should you film to get the best swap?

One person. Phone on a stand or propped against something, not handheld. Light on your face from the front or the side, not behind you. Vertical frame from the chest up or the waist up, with your head inside the frame the whole time. Talk at normal speed. Keep your hands below your chin most of the time and away from your face.

Avoid: dance, arm swings, jumping, walking toward the camera fast, hats and sunglasses, singing close-ups, other people in the frame, and very dark rooms. If your video needs a prop, hold it at chest height, not next to your face.

Length is not a problem. Clips up to 10 minutes are accepted, and each started 15 seconds uses credits, so a 2-minute video costs four times a 30-second one at the same level. Most people get the best results and the best price by cutting the clip to the parts they will post before uploading.

Step by step

1. Film a steady, well-lit clip of yourself
One person, phone on a stand, light on your face, vertical, chest up or waist up. Talk normally, hands below the chin. Up to 10 minutes.
2. Sign up and upload
Sign-up is free and needs no card. Upload the clip. The app checks the footage and greys out any level that would fail on it.
3. Pick a Stand-in, an outfit and a voice
Choose one of 16 ready-made Stand-ins or create your own. Keep what you wore or pick one of ten outfits. Keep your voice or swap it in one tap.
4. Pick the level and pay with credits
Lite, Standard or Pro. The video costs from $3.99 per 30 seconds (Lite, with a $9.37 credit pack); 1 credit is 1 cent and a $10 pack buys 1,000 credits.
5. Wait about 10 minutes and check the whole video
Watch it at full size. Check the mouth, the hands and the hairline. If something is wrong, it is redone free or the credits come back.
6. Post with the AI label on
TikTok: AI-generated toggle. YouTube: the AI use question under Attributes. Instagram: AI disclosure on the post. The platforms say the label alone does not cut reach.

Swap yourself in your own video

Questions people ask

Can I put my friend's face or a celebrity's face on my video?: Not with Stand-in, and you should not anywhere. Putting a real person's face on a video without consent is what deepfake laws target: the US TAKE IT DOWN Act (May 2025) and Denmark's proposed likeness rules are two examples. Every Stand-in is a generated person who belongs to no one.
How much does an AI face swap cost per 30 seconds?: Stand-in is from $3.99 per 30 seconds (Lite, with a $9.37 credit pack). Higgsfield's Seedance 2.5 route is about $10-$16 per 30 seconds at its October 2026 credit prices. Kling's Standard plan works out to about $3-$4 per 30 seconds of generated clips. HeyGen's face swap is on plans from $24 a month and takes clips up to 15 seconds.
Does the swap change my voice?: Only if you ask. Your recording's audio stays as it is by default, because the swap changes pixels, not sound. Stand-in lets you keep your voice or swap it in one tap at upload. Close friends may still recognize a kept voice; strangers will not.
Is this a deepfake?: A deepfake, in the way the law uses the word, puts a real person's likeness on something they did not do. A performance swap puts a fictional person onto something you did do, in your own footage, with a label on it. Same technology family, opposite purpose.
Why does my swap look wrong around the hands and mouth?: Hands crossing the face and fast mouth shapes are the two hardest things for any swap engine. Refilm with hands below the chin and talk at normal speed, or pick the Standard or Pro level. If a Stand-in video comes out wrong, it is redone free or the credits come back.
Do I need an AI label if only the face changed?: Yes. YouTube, TikTok and Instagram all ask for a label on realistic AI-altered video of people. The size of the edit is not the test; whether the result looks real is. YouTube exempts only minor edits like filters and color changes. Turn the label on when you post.
Can I swap a video that is already posted?: You can download your own video, swap it and repost it. Note that Instagram's AI label is set while the post is a draft, so the swapped version needs to be a new post, not an edit. On YouTube and TikTok you can also set the label on a fresh upload.
How long does a swap take?: Stand-in videos are usually ready in about 10 minutes: roughly 4 minutes of render per piece on Lite, 6 on Standard, 7 on Pro. Longer clips take longer. Prompt-based tools take a few minutes per 5-15 second clip plus the time to write and retry prompts.

What it costs with Stand-in

from $3.99 per 30-second video with a $9.37 pack (from $3.63 on Pro); from $2.00 up to 15 seconds. Free to join, no card. Packs from $9.37, and a first pack bought within 24 hours of sign-up comes with a free 30-second video. Up to 10 minutes on any account, priced in steps of 15 seconds. Your balance never expires, unused plan money rolls over, a failed video gives its price back, and a video that comes out wrong is redone free. Ready-made Stand-ins are free on every video.