How to Make a Self-Insert Video

A self-insert video puts you and your character in the same frame — a fandom staple usually built by hand-editing a photo of yourself next to fan art, frozen and static. LiveGen makes the live version: your character appears in your actual camera feed next to you, reacting in real time, so the "you and them" shot is something you film, not something you composite.

There are two routes to the insert, and this guide covers both. Route A: become the self-insert — use Face Swap so the person on camera is your sona or OC, moving with your real expressions. Route B: stay yourself and use Summon to bring the character into the frame beside you. The best clips often combine the two, and if the whole idea is new to you, self-shipping is the fandom shorthand for what you're about to make.

What you actually need

The list is short, and that's the point — this is a filming method, not an editing project.

• ✓A browser — desktop or mobile, no app, no install, no headset, no GPU on your end.
• ✓A camera — the webcam or phone camera you already have.
• ✓One reference image of your character (or a preset from the library), plus a face reference for your sona if you're taking Route A.
• ✓Nothing else. The heavy work runs in the cloud over WebRTC on the Xmax X2.0 model, frame by frame, so there's no render wait between takes and no editing software for the single-shot version.

Step-by-step

Gear & setup

No studio required, but a few choices separate a shot that reads as one scene from one that reads as a sticker on video:

• ✓Light both halves of the frame evenly. You're staging a two-character scene, so the light on you and the light on the character's landing spot need to match — one bright side and one dim side looks like a bad composite.
• ✓Lock the camera down. A tripod, phone stand, or stack of books. This matters double for two-pass skits, where any camera drift between takes breaks the illusion of a single scene.
• ✓Shoot at eye level. Self-insert scenes are conversations; a camera at sitting- or standing-eye height makes the two of you share the space, while a steep angle makes the character float.
• ✓Clear the landing surface. An uncluttered patch of desk or floor gives the character somewhere believable to be — and gives your summoning gesture room to read.
• ✓Use a stable connection. The feed runs live over WebRTC, so solid Wi-Fi or wired internet keeps the character's reactions in step with your movement.
• ✓Record clean audio if you're speaking. For dialogue formats, your voice is half the scene — a quiet room and a decent mic sell it more than any visual polish.

Scene formats worth stealing

Once the pipeline works, the question becomes what to stage. Formats that hold up:

• ✓Single-shot reaction duets. You and the summoned character react to the same thing — a trailer, a plot twist, a package. One take, no editing, done.
• ✓Dialogue skits in two passes. Lock the camera, then shoot the scene twice: pass one as yourself on one side of the frame, pass two face-swapped as your sona delivering the other half from the other side. Cut between passes on the dialogue turns and it plays as one conversation. Know the seams: the summoned character reacts to gestures, it doesn't deliver scripted lines — your voice-over, captions, or the swapped pass carries the dialogue.
• ✓OC-meets-canon clips. Face Swap into your original character, then summon the canon character into frame — the crossover shot fan artists spend hours compositing, filmed live instead.
• ✓POV self-ship moments. A greeting at the door, a coffee handoff, sitting together at the desk. Small, specific beats outperform grand ones; see Self-Ship Video for more of these.
• ✓Short-form hooks. A character popping in beside you mid-sentence is a strong opening beat for TikTok video effects and Reels.

Tips & common mistakes

• ✓Leave room in frame for both of you. Standing too close to the camera crowds out the space where the character appears.
• ✓Keep lighting consistent across you and the surface they're summoned onto, so the shot reads as one scene.
• ✓Match lighting and framing between passes. For two-pass skits, shoot back-to-back with the camera untouched and the lights unchanged — a nudged tripod or shifted lamp is what makes a cut feel like two different videos.
• ✓Give each character their own side of the frame and angle yourself slightly toward the other side, so eyelines meet when you cut between passes.
• ✓This is a live 2D composite, not physical contact — plan gestures that don't require touching them.
• ✓Keep clips short. A single clear moment (a greeting, a reaction) shares better than a long take.
• ✓Reshoot freely — there's no render wait, so try a few takes and keep the best one.

Do it with LiveGen

Summon places your character into the same live camera feed as you, reacting to your gestures in real time — the core of any self-insert video. Pair it with Face Swap if you also want to transform how you look in the shot. For more on the idea behind sharing a frame with a character, see Comfort Character and Anime Boyfriend, IRL.

Both routes run on the same real-time engine, so you can switch mid-session — shoot a Summon pass, flip to Face Swap for the sona pass, and never wait on a render between the two. Free credits cover your first experiments; free exports carry a watermark, and paid tiers remove it (verify current pricing). If the character-in-your-room half of the setup needs more depth, How to Put a Virtual Character in Your Room covers surfaces, gestures, and lighting in detail.