Qwen/Qwen-Image-2.1：7B 参数的统一文生图与图像编辑模型
英文标题：🤗 Qwen/Qwen-Image-2.1
Qwen 开源了 Qwen-Image-2.1，这是一款统一支持文本生成图像与图像编辑的模型，其视觉生成组件包含 70 亿参数和 32 个 Single-Stream DiT 层。该模型通过混合粒度注意力与 Prefix KV Cache 复用，在降低计算成本的同时兼顾图像质量、推理效率和多功能性。它原生支持透明 RGBA 图像生成与编辑、最多 10 张参考图、局部圈选或蒙版编辑，以及人物和商品身份保持。对于开发者而言，模型已提供基于 diffusers 的文生图、图像编辑、透明图生成及 CPU 卸载优化示例，可直接集成到生成式视觉应用中。
打开原文相关 1.6 · 机会 2
4.5 KB
2.8 KB
197.0 KB
290.4 KB
399.6 KB
967.7 KB
222.0 KB
4.1 MB
3.1 MB
3.1 MB
403.4 KB
中文
🤖 ModelScope | 🤗 HuggingFace | 📑 博客 | 🖥️ 演示 | 🫨 Discord
简介
我们很高兴开源 Qwen-Image-2.1。这是 Qwen 系列中一款统一的文本生成图像和图像编辑模型。其视觉生成组件仅包含 70 亿参数（32 个 Single-Stream DiT 层），在生成质量、推理效率和通用性之间取得平衡。
本次发布的四项关键改进包括：
- 紧凑且高效 —— 采用轻量级架构，结合混合粒度注意力和 Prefix KV Cache 复用，以较低的计算成本提供出色的图像质量。
- 原生透明支持，创作与编辑统一 —— 可根据文本生成普通图像或透明（RGBA）图像，编辑透明图层，并从照片中提取主体，所有能力均集成于同一个模型。
- 多功能编辑 —— 最多支持 10 张参考图像，可通过圆圈、绘制标注或独立蒙版指定局部编辑区域，并保持人物和商品的身份一致性。
- 真实纹理与精细美学 —— 改进排版文字、人像光照和精细细节，带来更具视觉吸引力的结果。
快速开始
安装
pip install torch>=2.4.0
pip install transformers>=5.17
pip install git+https://github.com/huggingface/diffusers
pip install accelerate pillow
文本生成图像
import torch
from diffusers import QwenImage21Pipeline
pipe = QwenImage21Pipeline.from_pretrained(
"Qwen/Qwen-Image-2.1", torch_dtype=torch.bfloat16
).to("cuda")
image = pipe(
prompt="A neon shop sign that reads \"QWEN IMAGE 2.1\", rainy night, reflections on wet pavement",
width=2048, height=2048,
num_inference_steps=40,
generator=torch.Generator("cuda").manual_seed(42),
).images[0]
image.save("t2i_example.png")
图像编辑
import torch
from PIL import Image
from diffusers import QwenImage21Pipeline
pipe = QwenImage21Pipeline.from_pretrained(
"Qwen/Qwen-Image-2.1", torch_dtype=torch.bfloat16
).to("cuda")
input_image = Image.open("input.png")
image = pipe(
prompt="Change the background to a sunset beach",
image=input_image,
num_inference_steps=40,
generator=torch.Generator("cuda").manual_seed(42),
).images[0]
image.save("edit_example.png")
透明图像生成（RGBA）
请使用推荐的透明图像提示词格式：
image = pipe(
prompt="This is an RGBA image with transparency. A cute cartoon dragon sticker. The image has alpha channel and the background is transparent.",
width=2048, height=2048,
num_inference_steps=40,
generator=torch.Generator("cuda").manual_seed(42),
).images[0]
image.save("transparent_example.png")
支持的宽高比
aspect_ratios = {
"1:1": (2048, 2048),
"4:3": (2400, 1792),
"3:4": (1792, 2400),
"3:2": (2528, 1696),
"2:3": (1696, 2528),
"16:9": (2752, 1536),
"9:16": (1536, 2752),
}
内存优化
pipe = QwenImage21Pipeline.from_pretrained(
"Qwen/Qwen-Image-2.1", torch_dtype=torch.bfloat16
)
pipe.enable_model_cpu_offload()
展示
原生透明图像生成
基于 6 张人像参考图生成的合影
文字渲染
许可证
该模型采用 Qwen Research License Agreement 授权。
English / 原文
🤖 ModelScope | 🤗 HuggingFace | 📑 Blog | 🖥️ Demo | 🫨 Discord
Introduction
We are excited to open-source Qwen-Image-2.1 , a unified text-to-image generation and image editing model in the Qwen family. With just 7B parameters in its visual generation component (32 Single-Stream DiT layers), Qwen-Image-2.1 balances generation quality, inference efficiency, and versatility.
Four key improvements define this release:
- Compact and Efficient — A lightweight architecture with mixed-granularity attention and prefix KV cache reuse delivers strong image quality at low computational cost.
- Native Transparency, Unified Creation and Editing — Generate regular or transparent (RGBA) images from text, edit transparent layers, and extract subjects from photographs—all in one model.
- Versatile Editing — Support up to 10 reference images , specify local edits via circles, painted annotations, or separate masks, and preserve identity for people and products.
- Realistic Textures and Refined Aesthetics — Improved typography, portrait lighting, and fine details for more visually compelling results.
For more details, see the GitHub repo and Blog.
Quick Start
Installation
pip install torch>=2.4.0
pip install transformers>=5.17
pip install git+https://github.com/huggingface/diffusers
pip install accelerate pillow
Text-to-Image
import torch
from diffusers import QwenImage21Pipeline
pipe = QwenImage21Pipeline.from_pretrained(
"Qwen/Qwen-Image-2.1", torch_dtype=torch.bfloat16
).to("cuda")
image = pipe(
prompt="A neon shop sign that reads \"QWEN IMAGE 2.1\", rainy night, reflections on wet pavement",
width=2048, height=2048,
num_inference_steps=40,
generator=torch.Generator("cuda").manual_seed(42),
).images[0]
image.save("t2i_example.png")
Image Editing
import torch
from PIL import Image
from diffusers import QwenImage21Pipeline
pipe = QwenImage21Pipeline.from_pretrained(
"Qwen/Qwen-Image-2.1", torch_dtype=torch.bfloat16
).to("cuda")
input_image = Image.open("input.png")
image = pipe(
prompt="Change the background to a sunset beach",
image=input_image,
num_inference_steps=40,
generator=torch.Generator("cuda").manual_seed(42),
).images[0]
image.save("edit_example.png")
Transparent Image Generation (RGBA)
Use the recommended prompt format for transparent images:
image = pipe(
prompt="This is an RGBA image with transparency. A cute cartoon dragon sticker. The image has alpha channel and the background is transparent.",
width=2048, height=2048,
num_inference_steps=40,
generator=torch.Generator("cuda").manual_seed(42),
).images[0]
image.save("transparent_example.png")
Supported Aspect Ratios
aspect_ratios = {
"1:1": (2048, 2048),
"4:3": (2400, 1792),
"3:4": (1792, 2400),
"3:2": (2528, 1696),
"2:3": (1696, 2528),
"16:9": (2752, 1536),
"9:16": (1536, 2752),
}
Memory Optimization
pipe = QwenImage21Pipeline.from_pretrained(
"Qwen/Qwen-Image-2.1", torch_dtype=torch.bfloat16
)
pipe.enable_model_cpu_offload()
Showcase
Native transparent image generation
Group photograph generated from six portrait references
Text rendering
License
This model is licensed under the Qwen Research License Agreement.