Erste Schritte

How to direct a silent reaction with Vidu Q4 Preview

Build a café listening shot around one emotional change, then check the eye line, mouth movement, and timing before cutting it into a conversation.

Start creating
How to direct a silent reaction with Vidu Q4 Preview

Vidu Q4:咖啡馆无声回应

Actual Vidu Q4 Preview generated sample.
Referenz hinzufügen (0/4)

Use the woman in input[0] as the character reference. Reframe to a close-up of her head and shoulders alone in the evening cafe; leave the conversation partner outside the frame, just left of the camera. She listens with her mouth closed, briefly lowers her eyes, then looks back toward the same eye line with a small, relieved smile. Keep the camera

Modell
Vidu Q4 Preview
Dauer
3s
Seitenverhältnis
16:9
Auflösung
540p

A conversation does not end when a character stops speaking. A glance toward a cup, a breath before answering, or a smile that arrives a moment late can carry the next cut. If every character keeps talking and moving, that small reaction disappears.

Start with a quiet café scene. The goal is a usable listening shot: one person receives information, pauses, then lets the expression change. This guide shows how to direct a silent reaction with Vidu Q4 Preview in ArcLoop.

A café reaction shot generated with Vidu Q4 Preview.

Frame extracted from the accompanying Vidu Q4 Preview sample. Watch the accompanying video example to judge the movement.

In the actual sample, the model reframes the two-person reference into a single listener close-up. Her gaze lowers and returns before a small smile develops, while the background stays steady. Her lips are slightly parted at the start, so the closed-mouth instruction was not fully followed. Use that detail as a review point when deciding whether the shot fits your dialogue edit.

Decide what the character has just heard

Write the emotional turn in plain language before touching the prompt. “She realises her friend remembers” gives the actor a direction. “Beautiful, emotional, cinematic” gives you no clear reason to accept or reject a take.

Keep the partner outside the frame for a single-person reaction. Choose where the partner sits relative to the lens and maintain that eye line. A look to camera can break the conversation even when the face itself looks convincing. If the reference shows two people, ask explicitly for a closer view of the intended listener in reference-to-video mode.

How to stage the reaction in ArcLoop

  1. Open your project and enter Canvas. Add a video node, select Vidu Q4 Preview, and attach an image that clearly shows the intended listener. Choose reference-to-video to reframe from that reference, or first-frame mode to retain an existing composition.

  2. Keep the scene readable. For this exercise, frame the head and shoulders, leave space toward the unseen partner, and keep hands out of the shot if they do not carry the emotion. Do not ask a close-up to solve a second action with a cup.

  3. Describe an emotional sequence with a pause: listening, looking down briefly, then returning the gaze. Name one final expression and keep the mouth closed. This separates a silent response from a speaking performance.

  4. Begin with a short clip. Our example uses three seconds at 540p with generated audio off. Check the estimate in the generator. These settings describe this sample, not a rule that every reaction needs three seconds.

  5. Review at normal speed before pausing individual frames. Keep a take only if the emotion reads without the written prompt. Record which eye line, expression, or timing needs a change before making another attempt.

Three reactions with different dramatic jobs

Replace the example asset names through ArcLoop's asset selector and attach the appropriate image. An @name typed as plain text does not attach a reference. The following descriptions are variations to try; the accompanying sample demonstrates one setup, not all three.

Recognition after a pause

Close-up of @Mira at @Evening Cafe, listening to a friend just left of the camera. She keeps her mouth closed, lowers her eyes for a brief pause, then looks back toward the same eye line with a small, relieved smile. Keep the camera locked and the warm window light steady. Frame only her head and shoulders so no hand movement competes with the expression. Let the smile arrive gradually rather than appearing immediately. Finish with the gaze settled on her friend, without a cut or spoken dialogue.

A guarded refusal

Close-up of @Mira at @Evening Cafe, facing an unseen conversation partner just right of the lens. She listens without smiling, takes a small breath, then gives one restrained shake of the head. Keep her mouth closed and her shoulders nearly still. Hold the camera at eye level throughout, preserving the warm side light and the soft background. The refusal should feel deliberate rather than angry. End with her gaze still on the partner and a quiet pause, without adding another gesture, a camera move, or a cut.

Holding back an answer

Medium close-up of @Mira at @Evening Cafe. Her attention stays on a partner just left of the lens. She begins to respond with a slight lift of the chin, stops herself before speaking, and settles into a thoughtful look. Keep the mouth closed after that small hesitation. The camera remains still, the lighting stays constant, and the background does not distract from the face. Leave a short moment of stillness at the end so the editor can cut back to the partner. No new person enters the frame.

Cut on the reaction, not on the prettiest frame

Watch where the eyes change direction and where the expression becomes readable. The beginning can serve as a listening hold; the ending can give the next speaker space. A technically smooth clip may still be unsuitable if the smile arrives before the implied line has landed.

Check blinking, mouth movement, and the direction of the eyes. Reject an unintended speaking motion if the shot must remain silent. Compare the face at the beginning and end, then check the largest expression change in the middle. Fix timing in the description; improve the reference if facial detail is already unclear.

Questions about silent performances

Does switching off audio stop the mouth moving?

No. Audio generation and visible performance are different controls. State the intended mouth behaviour in the prompt and inspect the result.

Should I add a moving camera for emotion?

Start with a locked camera so you can judge the performance. If the reaction works, try a separate version with a restrained push-in and compare which one serves the scene.

Can I guarantee the same expression every time?

No. Keep a clear reference and specific direction, then select a take that fits the surrounding shots. A prompt describes intent; it does not guarantee the exact timing of an expression.

Put the reaction between two dialogue shots

Choose one line from your scene, make a silent response to it, and watch the three shots together. The useful take is the one that makes the conversation understandable without extra explanation.

Try the café reaction template

Teilen

Put the reaction back into the conversation

Choose one line, make a silent response, and watch it between the two dialogue shots.

Start creating

Entdecken Sie mehr

AI Anime Drama Tools: A Capability Checklist for Where You're Stuck

AI Anime Drama Tools: A Capability Checklist for Where You're Stuck

Seed Audio 1.0 vs ElevenLabs for AI Drama Dubbing: Which Model to Use

Seed Audio 1.0 vs ElevenLabs for AI Drama Dubbing: Which Model to Use

Anime OC Maker for AI Short Series

Anime OC Maker for AI Short Series

Planting a Setup and Paying It Off

Planting a Setup and Paying It Off