Target Guide: AI video dialogue prompt

Two-Person Dialogue Scenes in AI Video: The Shot-Reverse-Shot Prompt Guide

Prompting two people having a conversation in a single video prompt is a recipe for disaster: characters talk at the same time or swap faces. Here is the Hollywood Shot-Reverse-Shot method tailored for AI models.

By Botock AI Filmmaking LabOctober 20, 202611 min readTested on Botock AI Video Generator

The Golden Rule of AI Dialogue

Never prompt both people speaking in the same clip. Hollywood films don't do this, and AI models can't handle it. Break dialogue into three distinct clips: 1) Speaker A delivers their line, 2) Speaker B reacts silently, 3) Speaker B replies.

The 3-Clip Dialogue Pipeline

Clip NumberShot TypeWho SpeaksWho Listens
Clip 01: Speaker DeliveryOver-the-shoulder on Person APerson A speaks exact linePerson B seen only from back of head/shoulder
Clip 02: Listener ReactionClose-up on Person BSilence (no speech)Person B reacts subtly with eyebrow raise or nod
Clip 03: The ResponseOver-the-shoulder on Person BPerson B speaks exact linePerson A seen only as foreground shoulder silhouette

Avoid vs. Correct: Production Prompt Comparison

AVOID (Glitches & Morphing)
"Two friends sitting at a diner talking to each other about their weekend plans and laughing."
Why it fails: Both characters will start murmuring simultaneously, faces will merge, and eyes will cross.
CORRECT (Production Ready)
CAMERA: Over-the-shoulder shot looking past Person B's blurred dark jacket at Person A.
SUBJECT: Person A (30s man, grey sweater) sits at frame-left facing frame-right.
ACTION: Person A speaks one line: 'I saw him yesterday.' He finishes speaking at 3 seconds and looks down.
AUDIO: Person A natural speaking voice. No background music.
Why it works: Anchors only one speaking face, uses classic cinema OTS framing, and specifies exact dialogue timing.

Production Step-by-Step Implementation

Step 1: Establishing Two-Shot

Start with a 3-second wide shot establishing both characters at the table without any dialogue.

Step 2: Shot-Reverse-Shot

Cut to over-the-shoulder for each spoken line.

Step 3: Capture the Silent Beat

A 2-second clip of pure facial reaction sells dramatic realism more than 10 seconds of talking.

Step 4: Edit in Botock Video Editor

Line up audio dialogue and reaction cuts on the multi-track timeline.

Interactive Studio

Test This Prompt in Botock Video Generator

Ready to generate your clip? Launch the Botock AI Video Generator right now with zero software installation and test these exact physical constraints.

Launch Video Generator