Remade-AI/Lego

Instala las dependencias y carga el LoRA sobre el modelo base Wan2.1 14B T2V:

pip install -U diffusers transformers accelerate

import torch
from diffusers import DiffusionPipeline
from diffusers.utils import export_to_video

# Cambia a "mps" para dispositivos Apple
pipe = DiffusionPipeline.from_pretrained(
    "Wan-AI/Wan2.1-T2V-14B",
    dtype=torch.bfloat16,
    device_map="cuda"
)
pipe.load_lora_weights("Remade-AI/Lego")

prompt = "l3g0_5ty13 Lego animation style, a close-up shot of a Lego pirate minifigure with a black tricorn hat, an eye patch, and a red coat. He is holding a silver cutlass in his right hand, and a small Lego parrot is perched on his shoulder. The background is a blurry mix of blue and brown Lego bricks resembling an ocean scene."

output = pipe(prompt=prompt).frames[0]
export_to_video(output, "output.mp4")

Para ComfyUI, utiliza el archivo lego_35_epochs.safetensors con el flujo wan_txt2vid_lora_workflow.json y conecta un nodo Wan LoRA al modelo base. Incluye al principio del prompt la frase de activación l3g0_5ty13 Lego animation style.