Writing Dialogue and Performance Beat Prompts in NadouPro

In NadouPro, you can guide character dialogue and performance by separating spoken lines, emotional beats, pauses, and shot context. For voice audio, use Text to Speech or Voice Dubbing together with performance settings. For the visual part, describe the character’s facial expression, body language, camera composition, and timing in the video or storyboard prompt instead of stuffing every instruction into the spoken dialogue.

NadouPro is iQIYI’s professional AI filmmaking and video creation platform. It combines Canvas, Workbench, Agents, Skills, and multimodal nodes, allowing creators to organize text, image, video, and audio assets in a project workflow from script and storyboard to final edit. You can start from NadouPro.

What should a dialogue prompt include?

A useful dialogue prompt usually has four layers:

• Line:The exact words the character should say.
• Delivery:Emotional or vocal direction, such as calm, frightened, bitter, excited, a low whisper, or slow speech.
• Beat:Pauses, reactions, breathing, eye movement, or changes in emotional state.
• Visual context:Shot size, camera angle, lighting, and what the character does in the frame.

Keeping these layers separate helps avoid a common mistake: writing a prompt that forces the model to guess which words are dialogue, which are performance instructions, and which are camera instructions.

A simple structure for character lines

For each line, write the prompt in a clean, production-oriented format:

Character:Maya, exhausted but restrained.
Line:“I told you not to open that door.”
Delivery:Low voice, suppressed anger, slow pace.
Beat:A brief pause before the final few words.
Visual:Close-up, eyes locked on the other character, dim key light from camera left.

This structure works well because the spoken content stays clear, while the performance and camera information are also easy to adjust in later iterations.

How to handle pauses and emotional beats

Pauses are not just silence. They tell the audience when a character is thinking, resisting, lying, realizing something, or changing their mind. In the visual prompt, describe what happens during the silence.

Beat type	Voice direction	Visual direction
Hesitation	Slower pace, brief pause before the key word	Eyes look away, breath held, shoulders tense
Shock	Very soft opening, then broken rhythm	Static close-up, face barely moves, pupils fixed
Controlled anger	Low volume, firm tone, restrained delivery	Jaw tightens, body stays still, hard key light
Relief	Softer tone, longer exhale, warmer pace	Shoulders relax, gaze softens, composition widens

Separate voice prompts from video prompts

If you are building a scene on Canvas, it is usually clearer to place dialogue or voice generation in an Audio node and place visual performance in Image, Video, Storyboard, or Clip-related workflows. NadouPro’s Canvas supports multimodal nodes and connections, and prompts can use @ to reference available assets. For example, when the selected node or model supports this input combination, a video prompt can reference an image for the character’s appearance and an audio asset for timing.

A practical way to split the work is as follows:

1. Write the scene beat:Define what changes emotionally in the scene.
2. Create the dialogue line:Keep the wording short and intentional.
3. Add voice direction:Use performance settings or tags to control emotion, pace, and pauses.
4. Design the shot:Specify shot size, angle, movement, and lighting separately.
5. Generate or refine assets:Iterate on audio, image, video, and editing nodes instead of forcing one prompt to solve everything.

NadouPro dialogue scene prompt template

Scene goal:[What this scene is emotionally about.]
Character state:[What the character wants, fears, or is hiding.]
Dialogue:“[Exact spoken line.]”
Voice performance:[Emotion, pace, volume, pauses, vocal action.]
Performance beats:[Face, gaze, gestures, body tension, reaction.]
Shot language:[Close-up, medium shot, over-the-shoulder shot, eye level, low angle, dolly in, locked-off shot, etc.]
Lighting and mood:[Key light, rim light, soft light, color tone, atmosphere.]

Use only the details that truly matter. Too many emotional adjectives can make the performance unclear. A focused direction such as “quiet, restrained anger, with one long pause before the final sentence” is usually more useful than a long list of mixed emotions.

Common mistakes to avoid

• Putting camera instructions into the dialogue.Keep spoken text separate from shot direction.
• Asking for conflicting emotions.If the character is both afraid and happy, explain the beat of the emotional transition.
• Over-tagging every phrase.Use tags at important turning points instead of tagging every word.
• Expecting one prompt to control every model in the same way.Different generation modes and models may follow performance details differently.
• Forgetting reaction shots.Performance is usually clearer when the prompt includes how the character reacts before and after speaking.

When to use Workbench instead of a single prompt

Use Workbench when the dialogue belongs to a larger scripted scene. NadouPro’s Workbench organizes projects around scripts, subjects, storyboards, and shot production. This is useful when you need recurring characters, multiple shots, or consistent story logic across a sequence. A single prompt is suitable for testing one line; Workbench is better for building a scene.

Next steps

Start with one short line, one emotion, one pause, and one shot type. Generate the voice and visual performance separately, then refine the timing and performance beats. Once the line works, expand the scene into storyboard shots, or connect the related text, audio, image, and video assets on Canvas.

FAQ

Can NadouPro make characters speak with emotion?

NadouPro materials describe Text to Speech or Voice Dubbing with performance settings; specific results may vary depending on the selected voice capability and generation context.

Should I write pauses in the script or in the voice prompt?

Use both carefully. Mark important pauses in the voice direction, or use supported pause tags, and describe the visual reaction during the pause in the video or storyboard prompt.

What is the best prompt style for a dialogue scene?

The most reliable structure is: character state, exact line, voice delivery, emotional beat, and shot context. This prevents dialogue, performance, and camera direction from competing with each other.

Contains AI-generated information that may be incorrect. Please verify independently.