On-screen dialogue

On-screen dialogue belongs to a visible speaker and a specific moment. Separate the literal words from delivery, gestures and camera direction. Russian terms: диалог, реплики.

Studio example

The gardener looks at the empty pot and says, “We need one more seed.” The courier listens without speaking, then points toward the greenhouse.

Put the speaker and exact words in the scene’s dialogue, then let Prompt for the model adapt the request to the selected family. Check the model’s available audio controls and declared speech languages. A separate Voiceover and Lip-sync path has different inputs and must be prepared explicitly.

Verify the result

Listen for every word, language and pronunciation, then check which mouth moves. Automatic transcription is supporting evidence, not a substitute for listening. The 11 October HappyHorse trial produced Russian speech with a mispronounced word; the Grok trial does not establish all-language reliability.

Availability and evidence

This technique is documented as an authoring or workflow practice. It is available only through the stated inputs and operations; it is not a new model parameter. An unrendered example is untested. A dated successful case is an observation for its exact model, version, quality and mode, not a guarantee for another scene. Choose a model by the required operation first.

Related guides: voice, lipsync, transcription, narration.

See the model–technique applicability matrix for connected operations, dated outcomes and configurations that remain untested.

Reviewed 14 October 2026 against the current Studio authoring contract and the evidence described above. No new paid render was made for this example.