§
    ÎÐjn  ã                   óV   — d dl mZmZ  eddddddg¦  «        Z edd	¦  «        ed
<   dS )é    )Ú	help_textÚspecialitieszMagiHuman Talking CharactersaÜ  Animate a **Start Image** into a human-centered video with speech and sound. Choose text-driven soundtrack generation, or supply **Driving Audio** so that its recording sets the words and timing. A stable portrait or upper-body composition is a good starting point; large body actions and moving cameras are less suited to this workflow. Sliding windows can extend a performance; inspect identity and voice continuity. Distilled and full models use their own sampling presets.u1  ### Performance, Dialogue and Sound

Write the visual description in **English**, using chronological facial, head and shoulder movements. For a detailed performance, aim for about **150â€“200 words** covering appearance, expression, vocal delivery and a stable camera. Keep the spoken words in their original language. Add **Dialogue:** with a speaker description and language, then **Background Sound:**.

**Short example:**

```text
A woman in a green sweater sits at a desk, facing a fixed camera. She smiles and calmly says "The parcel is ready." Soft daylight, subtle head movement.

Dialogue:
Woman wearing a green sweater, English: "The parcel is ready."

Background Sound:
Quiet room tone.
```

For supplied Driving Audio, match the description to that performance instead of asking for conflicting dialogue.zèStart image + prompt generate synchronized video/audio, or audio_prompt_type A uses audio_guide as driving soundtrack. Best with a stable portrait and small expressive movements. Sliding windows extend performance; check continuity.uý   Use an English visual performance paragraph (about 150â€“200 words for detail), then `Dialogue:` with `speaker description, language: "exact words"` in the spoken language, then `Background Sound:`. Keep camera/body stable; match supplied driving audio.z/https://github.com/SandAI-org/daVinci-MagiHumanzaudio-video generationzlip syncr   N)Úmodels.help_sharedr   r   Ú	MAGI_HELP© ó    ú?C:\Users\johnn\OneDrive\Documents\AI\models\magi_human\infos.pyú<module>r
      s{   ðà 6Ð 6Ð 6Ð 6Ð 6Ð 6Ð 6Ð 6àˆIÐ4ð  7Uð Wuð  wað  cb	ð  e	V
ð  d	W
ñ!X
ô X
€	ð" )˜LÐ)AÀ:ÑNÔN€	ˆ.Ñ Ð Ð r   