Qwen-Image-Layered

image-layered.py

import torch
import torch.nn.functional as F

_orig_sdp = F.scaled_dot_product_attention

def _safe_sdp(*args, **kwargs):
    # diffusers 側が query/key/value で渡してくるケースに対応
    if len(args) >= 3:
        q, k, v = args[:3]
        rest_args = args[3:]
    else:
        # どっちの名前でも拾う
        q = kwargs.pop("q", kwargs.pop("query", None))
        k = kwargs.pop("k", kwargs.pop("key", None))
        v = kwargs.pop("v", kwargs.pop("value", None))
        if q is None or k is None or v is None:
            raise TypeError("_safe_sdp: missing q/k/v (or query/key/value)")
        rest_args = ()

    try:
        return _orig_sdp(q, k, v, *rest_args, **kwargs)
    except TypeError as e:
        # enable_gqa が未対応の torch でも落ちないように保険
        if "enable_gqa" in kwargs:
            kwargs.pop("enable_gqa", None)
            return _orig_sdp(q, k, v, *rest_args, **kwargs)
        raise

F.scaled_dot_product_attention = _safe_sdp



from diffusers import QwenImageLayeredPipeline
import torch
from PIL import Image

pipeline = QwenImageLayeredPipeline.from_pretrained("Qwen/Qwen-Image-Layered")
pipeline = pipeline.to("cuda", torch.bfloat16)
pipeline.set_progress_bar_config(disable=None)

image = Image.open("test.png").convert("RGBA")
inputs = {
    "image": image,
    "generator": torch.Generator(device='cuda').manual_seed(777),
    "true_cfg_scale": 4.0,
    "negative_prompt": " ",
    "num_inference_steps": 50,
    "num_images_per_prompt": 1,
    "layers": 4,
    "resolution": 640,      # Using different bucket (640, 1024) to determine the resolution. For this version, 640 is recommended
    "cfg_normalize": True,  # Whether enable cfg normalization.
    "use_en_prompt": True,  # Automatic caption language if user does not provide caption
}

with torch.inference_mode():
    output = pipeline(**inputs)
    output_image = output.images[0]

for i, image in enumerate(output_image):
    image.save(f"{i}.png")