AI Scene Builder: upload two character portraits, pick a cinematic preset, and generate a two-character video.
Or pick a sample portrait:
Or pick a sample portrait:
0 / 4000
Add characters, pick a scene preset, and modify to your liking.
AI Scene Builder example
Two sample portraits and the Wedding preset — the same inputs used to generate the clip below.
Input
Character 1 reference portrait
Character 2 reference portrait
Scene preset
Wedding
The Wedding preset auto-fills a multi-shot cinematic prompt with ceremony beats, dialogue, and camera directions.
Output
Generated with VEO 3 Fast reference image to video (~8 seconds, 16:9).
Use the sample portraits under each upload slot to recreate this AI Scene Builder example, or swap in your own characters.
Build two-character videos with AI Scene Builder
Use it when you need both faces to stay recognizable from the wide shot through the close-up, without restarting the prompt for every cut.
AI Scene Builder is Pixwit’s two-character video workspace: you upload two reference portraits, choose a cinematic preset, and generate a multi-shot clip without writing a shot list from scratch. Wedding vows, a dance battle, a late-night drive, a dinner date — each preset already carries camera beats, dialogue cues, and pacing notes you can keep or rewrite.
The form on this page is the product. Add Character 1 and Character 2, pick Boxing Match, Wedding, Dinner Date, Dance Battle, Kissing, Mafia Meeting, Late Night Drive, Beach Episode, Morning Together, or Custom, then generate an approximately eight-second 16:9 clip. First runs can use signup credits; each generation costs 40 credits, the same as Pixwit’s reference-image-to-video path.
Two-Character Video
How to use AI Scene Builder
Four steps from two portraits to a cinematic two-character clip
Upload two character portraits
Add a JPEG or PNG for Character 1 and Character 2. Clear, front-facing photos with even light work best; both slots are required before you generate.
Pick an AI Scene Builder preset
Choose one of nine cinematic templates — boxing, wedding, dinner date, dance battle, and more — or switch to Custom if you already know the scene.
Edit the cinematic prompt
Every preset fills a multi-shot prompt with establishing shots, close-ups, and dialogue beats. Rewrite any line, or start from a blank Custom brief.
Generate the two-character video
Click Generate to build an ~8 second clip that uses both portraits as identity locks. Download the file when the job finishes.
What an AI Scene Builder is
A guided two-character video tool — not a blank text-to-video box
An AI Scene Builder is a guided generator that turns two still portraits into a short scene with shared blocking, camera language, and (when the preset includes it) spoken beats. You are not hoping a single paragraph remembers both faces, wardrobe, and a shot list. The portraits lock identity; the preset supplies structure; the prompt stays editable so you can change a line of dialogue or a camera move without rebuilding the whole brief.
People searching this topic usually want one of three things: a 3D layout tool, a storyboard app, or a video generator that can put two specific people in the same frame. Pixwit is the third kind. It does not export a mesh or a lighting diagram. It produces a downloadable cinematic clip you can post, cut into a longer edit, or use as a reference for the next take.
Two-character work is where unconstrained prompting often drifts: the second person swaps outfits, the wide shot ignores the close-up identity, or the model invents a third extra. Feeding two references and a preset with named beats is how you keep a couple, a host and guest, or two opponents looking like the same pair from the first frame to the last. That is the job this page is built for.
Why creators use AI Scene Builder
Dual portrait locks, nine cinematic templates, and a prompt you can still rewrite
Two-character identity lock
Both portraits guide faces, hair, and wardrobe so Character 1 and Character 2 stay distinct across establishing shots and close-ups instead of melting into one generic extra.
Nine cinematic scene presets
Wedding, dance battle, boxing, dinner date, late-night drive, beach, morning together, mafia meeting, and kissing — each ships with shot order and style notes, not a one-line slogan.
Fully editable multi-shot prompts
Presets are starting scripts, not locks. Change a toast, a camera height, or a location, or ignore the template and write a Custom scene in your own words.
Reference-image video engine
Generation uses Pixwit’s VEO 3 Fast reference-to-video path, tuned for motion that still respects the stills you uploaded rather than inventing new faces mid-clip.
Samples you can try before you upload
Each slot includes portrait samples so you can test a preset, see the wedding example below, then swap in your own characters when the blocking looks right.
Who an AI Scene Builder is for
Couple films, dialogue shorts, and campaign talent that has to stay on-model
Couple and wedding films
Turn two portraits into a ceremony walk, a first look, or a beach vow without hiring a second unit. Keep the pair recognizable when you cut from wide to close-up.
Dialogue and interview shorts
Host-and-guest or rival-and-rival scenes need two stable faces. Use a dinner, mafia meeting, or boxing preset, then rewrite the spoken beats to match your script.
Social and campaign talent
Lock two spokespersons once, generate a short interaction, and reuse the same stills in later ads so the cast does not change between posts.
Story exploration before a full shoot
Try a dance battle versus a late-night drive with the same two people. Presets are cheaper than a location scout when you are still choosing the mood.
How AI Scene Builder fits a two-character workflow
Treat the two portraits as casting, the preset as a shot list, and the prompt as the rewrite pass. Generate, watch whether both identities hold, then change one beat — a line, a location, a camera height — instead of throwing the whole brief away. Eight-second clips are building blocks: stitch a few together in an editor, or pick the strongest take and post it as a standalone scene.
If you still need a matching full-body still before you animate, build the person in Character Builder and upload that portrait into a character slot here.
For a single subject, or when you want more reference images without a two-person preset, use Reference Image to Video on the same engine. Scene Builder is the guided two-character lane; the reference tool is the open lane.
When you rewrite camera language, think in coverage: a wide to establish geography, a two-shot for relationship, then singles for reaction. That is standard cinematography coverage— the presets encode a short version of it so you are not inventing a shot list under a blank prompt.