Segmind Charm Collider

Segmind Charm Collider is an AI video face and body swap that puts the person from one photo into a video clip. Give it a clip, a photo and a few words saying who to replace. It returns the same take with the new person in it: their face, hairstyle, skin tone and facial hair, with every movement, expression, lip movement and camera move from the original. The background and everyone else stay as they were. Then it finishes the take for posting or editing: your clip's exact length and frame rate, smooth motion, its grain and its sound.

Put yourself in any video with one photo

For creators, fans and anyone who wants to be in the shot. Bring a clip and a selfie.

• •Star in a cinematic scene: take the lead in a dramatic close-up, a chase or a trailer-style shot you have the rights to.
• •Trends, memes and reactions: join a trending dance, skit or reaction format with your own face, on the original beat.
• •Surprise friends and family: put a friend, with their OK, into a birthday, wedding or farewell video moment.
• •One take, many faces: film a performance once, then put each member of your crew or series cast in it.

Works best with:

• •a clear, front-facing photo cropped to the head;
• •a clip where the face is easy to see;
• •1080p for close-ups and small faces.

Recast an actor in a shot without a reshoot

For film, TV and ad studios. The performance, camera and set stay; the person changes.

• •Recasting and pickups: replace an actor or an extra after the shoot has wrapped, keeping the take you already approved.
• •Ad variants for every market: turn one licensed performance into versions with different on-screen talent, each with the talent's consent.
• •Stunt doubles and stand-ins: put the lead's face on the double or stand-in, so the shot cuts with the rest of the scene.
• •Casting tests and pitches: see how a candidate reads in the role, or pitch a scene with the intended actor's look before the shoot.

What stays the same:

• •every movement, expression and lip movement;
• •the framing and the camera move;
• •the clothes from the clip, unless you pick the photo's;
• •the background and everyone else in the shot.

How to use it

1. •Upload your clip: 4 to 30 seconds, one continuous shot.
2. •Upload a photo of the person to put in it: a clear, front-facing photo cropped to the head works best. Clothes that show in the photo can come through.
3. •Say who to replace in Who, for example "the young man with dark hair" or "the woman in the red coat".
4. •Pick the outfit. The default keeps the clothes from the clip; "wear the clothes from the photo" dresses them as in the photo.
5. •Pick a resolution: leave it at 720p, pick 1080p for close-ups, or 480p for a quick test.
6. •Run.

Output

• •Resolution: 720p by default, or 480p or 1080p.
• •Format: MP4 (H.264, 8-bit) with the clip's audio as AAC, so it plays in every browser and on phones.
• •Length and frame rate: exactly your clip's. The take is stretched back with smooth optical-flow in-betweens rather than repeated frames.
• •Sound: the clip's own audio is put back, so the dialogue, music and lip sync line up.
• •Grain match: the clip's grain is measured and added back, so the new face sits in the footage.
• •People per run: one.

Use your own clips and faces

Use clips you own or have the rights to, and a photo of yourself or of someone who has given permission. Don't use Charm Collider to impersonate anyone or to make content that misleads.

The example clip is from Tears of Steel (c) Blender Foundation, mango.blender.org, licensed under CC BY 3.0. The actor was replaced with a generated, fictional person.