Avatar Video Endpoint

Generate an avatar (talking-head) video using VEED Fabric 1.0.

Turns a portrait/face image into a lip-synced talking video. Audio can be supplied directly via audio_url or generated on-the-fly from voiceover_text using TTS (Hume / ElevenLabs, depending on the voice's provider).

Args: request: JSON body containing: - project_id (str, required): Project ID - image_url (str, optional): URL of the avatar/face image - start_keyframe_id (str, optional): Keyframe ID to use as avatar image (resolves image_url) - voiceover_text (str, optional): Script text for TTS generation - audio_url (str, optional): URL of pre-recorded audio for lip-sync - voice_id (str, optional): Voice ID for TTS (project voice record) - model (str, optional): "standard" or "fast". Default: "standard" - resolution (str, optional): "480p" or "720p". Default: "720p" - video_id (str, optional): Existing video ID to regenerate in place - order_index (int, optional): Position in timeline. Default: append at end