config.py

"""
Configuration file for WAN-VACE video generation application
"""
import os

# Hugging Face token (must be set as environment variable)
HF_TOKEN = os.getenv("HF_TOKEN")

# Model paths and configurations
MODEL_CONFIG = {
    "transformer_path": "https://huggingface.co/calcuis/wan-gguf/blob/main/wan2.1-v5-vace-1.3b-q4_0.gguf",
    "text_encoder_path": "chatpig/umt5xxl-encoder-gguf",
    "text_encoder_file": "umt5xxl-encoder-q4_0.gguf",
    "vae_path": "callgg/wan-decoder",
    "pipeline_path": "callgg/wan-decoder"
}

# Default generation parameters
DEFAULT_PARAMS = {
    "width": 720,
    "height": 480,
    "num_frames": 57,
    "num_inference_steps": 24,
    "guidance_scale": 2.5,
    "conditioning_scale": 0.0,
    "fps": 16,
    "flow_shift": 3.0
}

# UI configuration
#
# The title and description here emphasise the agentic nature of the app:
# you provide a concept and the system plans the prompts for you.  Feel free
# to adjust these strings to suit your needs or branding.
UI_CONFIG = {
    "title": "🎬 Agentic WAN-VACE Video Generation",
    "description": (
        "Generate high-quality videos from simple concepts. "
        "Provide a short description of what you want to see, and the agent "
        "will craft a refined prompt and negative prompt before generating a cinematic "
        "vertical video using the WAN‑VACE model."
    ),
    "theme": "default"
}

# Server configuration
SERVER_CONFIG = {
    "host": "0.0.0.0",
    "port": 7860,
    "share": False
}