Avatar Video Pipeline

Fresh

Overview

Creatify offers three avatar video models: Lipsync v1 (single scene), Lipsync v2 (multi-scene), and Aurora (SOTA from image). This workflow helps you choose the right model and execute the full pipeline.

Decision Tree

Model Comparison

Feature	Lipsync v1	Lipsync v2	Aurora
Endpoint	`/api/lipsyncs/`	`/api/lipsync_v2/`	`/api/aurora/`
Input	Avatar ID + text/audio	Scene array	Image + text/audio
Scenes	1	Multiple	1
Preview + Render	No	Yes	No
Quality	Good	Good	Studio-grade
Speed	Fast	Medium	Slower
Credits (create)	5/30s	5/30s	SOTA pricing
Credits (render)	N/A	4/30s	N/A
Best for	Quick avatar videos	Narratives, demos	Premium/branded content

Pipeline 1: Lipsync v1 (Single Scene)

python

# Lipsync v1 flow
avatar_id = "avatar-uuid"
voice_id = "voice-uuid"

task = requests.post(
    "https://api.creatify.ai/api/lipsyncs/",
    headers=headers,
    json={
        "text": "Your script here.",
        "creator": avatar_id,
        "voice_id": voice_id,
        "aspect_ratio": "9:16"
    }
).json()

result = poll_until_done(
    f"https://api.creatify.ai/api/lipsyncs/{task['id']}/",
    headers
)
print(f"Video: {result['output']}")

Pipeline 2: Lipsync v2 (Multi-Scene)

python

# Lipsync v2 flow
task = requests.post(
    "https://api.creatify.ai/api/lipsync_v2/",
    headers=headers,
    json={
        "scenes": [
            {
                "text": "Scene 1: Introduction",
                "creator": "avatar-1",
                "voice_id": "voice-1",
                "background": "office"
            },
            {
                "text": "Scene 2: Features",
                "creator": "avatar-2",
                "voice_id": "voice-2",
                "background": "studio"
            }
        ],
        "aspect_ratio": "9:16"
    }
).json()

# Poll preview
preview = poll_until_done(
    f"https://api.creatify.ai/api/lipsync_v2/{task['id']}/",
    headers
)

# Render
requests.post(
    f"https://api.creatify.ai/api/lipsync_v2/{task['id']}/render/",
    headers=headers
)

# Poll render
final = poll_until_done(
    f"https://api.creatify.ai/api/lipsync_v2/{task['id']}/",
    headers
)
print(f"Video: {final['output']}")

Pipeline 3: Aurora (SOTA from Image)

Error Handling

Error	v1 Action	v2 Action	Aurora Action
Invalid avatar	Re-fetch `/personas/`	Check each scene's creator	N/A (uses image)
Invalid voice	Re-fetch `/voices/`	Check each scene's voice_id	Re-fetch `/voices/`
Failed status	Check `failed_reason`	Check `failed_reason`	Check `failed_reason`
Low quality	Try different avatar	Try different avatars/backgrounds	Use higher-res image

Avatar Video Pipeline ​

Overview ​

Decision Tree ​

Model Comparison ​

Pipeline 1: Lipsync v1 (Single Scene) ​

Pipeline 2: Lipsync v2 (Multi-Scene) ​

Pipeline 3: Aurora (SOTA from Image) ​

Error Handling ​