Update website with Qwen-Image-Edit content and WaveSpeed links

README.md

@@ -1,11 +1,15 @@
 ---
 title: Qwen-Image-Edit
-emoji: 💻
+emoji: 🖼️
 colorFrom: blue
 colorTo: gray
 sdk: static
 pinned: false
-short_description: Qwen-Image-Edit
+short_description: Advanced image editing foundation model by Alibaba Cloud
 ---
 
-Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
+# Qwen-Image-Edit
+
+Qwen-Image-Edit is a powerful image editing foundation model in the Qwen series that achieves significant advances in transforming existing images with precise control. Experiments show strong capabilities in image-to-image generation, with exceptional performance in maintaining original image structure while applying creative transformations.
+
+Check out the model on Hugging Face: https://huggingface.co/Qwen/Qwen-Image-Edit

index.html

@@ -11,7 +11,8 @@
     <meta property="og:title" content="Qwen-Image Edit - Advanced Image-to-Image Generation by Alibaba Cloud" />
     <meta property="og:description" content="Transform your existing images into stunning new creations with Qwen-Image Edit, part of the Tongyi Qianwen model series developed by Alibaba Cloud" />
     <meta property="og:type" content="website" />
-    <meta property="og:url" content="https://huggingface.co/Qwen/Qwen-Image-Edit" />
+    <meta property="og:url" content="https://wavespeed.ai/models/wavespeed-ai/qwen-image/edit" />
+    <meta property="og:image" content="https://qianwen-res.oss-cn-beijing.aliyuncs.com/Qwen-Image/merge3.jpg" />
     
     <!-- Additional Meta Information -->
     <meta name="author" content="Alibaba Cloud Qwen Team" />
@@ -25,10 +26,10 @@
         <div class="nav-content">
             <div class="nav-logo">QWEN</div>
             <div class="nav-links">
-                <a href="https://wavespeed.ai/models/wavespeed-ai/qwen-image/edit" class="nav-link" target="_blank" rel="noopener noreferrer">Home</a>
-                <a href="https://wavespeed.ai/collections/qwen" class="nav-link" target="_blank" rel="noopener noreferrer">Documentation</a>
-                <a href="https://wavespeed.ai/collections/qwen" class="nav-link" target="_blank" rel="noopener noreferrer">Blog</a>
-                <a href="https://wavespeed.ai/models/wavespeed-ai/qwen-image/edit" class="nav-button" target="_blank" rel="noopener noreferrer">Visit WaveSpeedAI →</a>
+                <a href="https://wavespeed.ai/" class="nav-link" target="_blank" rel="noopener noreferrer">Home</a>
+                <a href="https://wavespeed.ai/models/wavespeed-ai/qwen-image" class="nav-link" target="_blank" rel="noopener noreferrer">Documentation</a>
+                <a href="https://wavespeed.ai/models/wavespeed-ai/qwen-image" class="nav-link" target="_blank" rel="noopener noreferrer">Blog</a>
+                <a href="https://wavespeed.ai/models/wavespeed-ai/qwen-image/edit" class="nav-button" target="_blank" rel="noopener noreferrer">Try on WaveSpeed →</a>
             </div>
         </div>
     </nav>
@@ -41,9 +42,9 @@
             </div>
             
             <div class="announcement-section">
-                <p class="announcement">Qwen-Image Edit is now available!</p>
+                <p class="announcement">Qwen-Image Edit is now open-source!</p>
                 <div class="divider"></div>
-                <p class="description">Open-source Advanced Image-to-Image Generative Model</p>
+                <p class="description">Advanced Image-to-Image Generative Model for Precise Editing</p>
             </div>
             
             <div class="hero-image">
@@ -52,7 +53,7 @@
 
             <section class="intro-section">
                 <h2>Introduction</h2>
-                <p>We are thrilled to release Qwen-Image Edit, an image editing foundation model in the Qwen series that achieves significant advances in transforming existing images with precise control. Experiments show strong capabilities in image-to-image generation, with exceptional performance in maintaining original image structure while applying creative transformations.</p>
+                <p>We are thrilled to release Qwen-Image Edit, an image editing foundation model in the Qwen series that achieves significant advances in transforming existing images with precise control. Experiments show strong capabilities in image-to-image generation, with exceptional performance in maintaining original image structure while applying creative transformations. Qwen-Image Edit is now available on Hugging Face and can be used locally with the diffusers library.</p>
                 <div class="benchmark-image">
                     <img src="https://qianwen-res.oss-cn-beijing.aliyuncs.com/Qwen-Image/bench.png" alt="Qwen-Image Benchmark" class="full-width-img">
                 </div>
@@ -78,7 +79,44 @@
                 <h2>Quick Start</h2>
                 <p>Choose your preferred Qwen image model:</p>
                 
-                <h3>Option 1: Using the latest Qwen VLo model</h3>
+                <h3>Option 1: Using Qwen-Image-Edit with diffusers</h3>
+                <p>Install the latest version of diffusers</p>
+                <div class="code-block">
+                    <pre><code>pip install git+https://github.com/huggingface/diffusers</code></pre>
+                </div>
+                <div class="code-block">
+                    <pre><code>import os
+from PIL import Image
+import torch
+from diffusers import QwenImageEditPipeline
+
+pipeline = QwenImageEditPipeline.from_pretrained("Qwen/Qwen-Image-Edit")
+print("pipeline loaded")
+pipeline.to(torch.bfloat16)
+pipeline.to("cuda")
+pipeline.set_progress_bar_config(disable=None)
+
+image = Image.open("./input.png").convert("RGB")
+prompt = "Change the rabbit's color to purple, with a flash light background."
+
+inputs = {
+    "image": image,
+    "prompt": prompt,
+    "generator": torch.manual_seed(0),
+    "true_cfg_scale": 4.0,
+    "negative_prompt": " ",
+    "num_inference_steps": 50,
+}
+
+with torch.inference_mode():
+    output = pipeline(**inputs)
+
+output_image = output.images[0]
+output_image.save("output_image_edit.png")
+print("image saved at", os.path.abspath("output_image_edit.png"))</code></pre>
+                </div>
+                
+                <h3>Option 2: Using the latest Qwen VLo model</h3>
                 <p>The new Qwen VLo model specializes in image-to-image generation with progressive editing features.</p>
                 <div class="code-block">
                     <pre><code>pip install dashscope>=1.20.7</code></pre>
@@ -109,66 +147,6 @@ if response.status_code == 200:
 else:
     print(f'Failed to generate image: {response.message}')</code></pre>
                 </div>
-                
-                <h3>Option 2: Using Qwen-Image Edit with diffusers</h3>
-                <p>Install the latest version of diffusers</p>
-                <div class="code-block">
-                    <pre><code>pip install git+https://github.com/huggingface/diffusers</code></pre>
-                </div>
-                <p>The following contains a code snippet illustrating how to use the model to generate images based on text prompts:</p>
-                <div class="code-block">
-                    <pre><code>from diffusers import DiffusionPipeline
-import torch
-
-model_name = "Qwen/Qwen-Image-Edit"
-
-# Load the pipeline
-if torch.cuda.is_available():
-    torch_dtype = torch.bfloat16
-    device = "cuda"
-else:
-    torch_dtype = torch.float32
-    device = "cpu"
-
-pipe = DiffusionPipeline.from_pretrained(model_name, torch_dtype=torch_dtype)
-pipe = pipe.to(device)
-
-positive_magic = {
-    "en": "Ultra HD, 4K, cinematic composition.", # for english prompt
-    "zh": "超清，4K，电影级构图" # for chinese prompt
-}
-
-# Load input image
-init_image = PIL.Image.open("input_image.jpg").convert("RGB")
-
-# Define editing prompt
-prompt = '''Transform this image into a watercolor painting with vibrant colors and artistic brush strokes. Ultra HD, 4K, cinematic composition'''
-
-negative_prompt = " "
-
-# Generate with different aspect ratios
-aspect_ratios = {
-    "1:1": (1328, 1328),
-    "16:9": (1664, 928),
-    "9:16": (928, 1664),
-    "4:3": (1472, 1140),
-    "3:4": (1140, 1472)
-}
-
-width, height = aspect_ratios["16:9"]
-
-image = pipe(
-    prompt=prompt + positive_magic["en"],
-    negative_prompt=negative_prompt,
-    image=init_image,  # Input image for editing
-    num_inference_steps=50,
-    strength=0.75,  # Control how much to transform the original image
-    true_cfg_scale=4.0,
-    generator=torch.Generator(device="cuda").manual_seed(42)
-).images[0]
-
-image.save("example.png")</code></pre>
-                </div>
             </section>
 
             <section class="showcase-section">
@@ -176,41 +154,51 @@ image.save("example.png")</code></pre>
                 
                 <div class="showcase-item-full">
                     <div class="showcase-description-full">
-                        <h3>Superior Image Transformation</h3>
-                        <p>One of its standout capabilities is high-fidelity image transformation across diverse styles and contexts. Qwen-Image Edit preserves the essential structure and content of the original image while applying sophisticated transformations, maintaining coherence and contextual harmony with stunning accuracy. The editing isn't just superficial—it's intelligently integrated into the visual fabric.</p>
+                        <h3>Semantic Editing</h3>
+                        <p>One of the highlights of Qwen-Image Edit lies in its powerful capabilities for semantic editing. It can modify image content while perfectly preserving the original visual semantics. For example, when editing character images like Qwen's mascot Capybara, the model maintains character consistency even when most pixels in the image are changed. This enables effortless and diverse creation of original IP content, such as MBTI-themed emoji packs based on mascot characters.</p>
                     </div>
                     <div class="showcase-image-full">
-                        <img src="https://qianwen-res.oss-cn-beijing.aliyuncs.com/Qwen-Image/s1.jpg" alt="Text Rendering Example" class="showcase-img-full">
+                        <img src="https://qianwen-res.oss-cn-beijing.aliyuncs.com/Qwen-Image/s1.jpg" alt="Semantic Editing Example" class="showcase-img-full">
                     </div>
                 </div>
 
                 <div class="showcase-item-full">
                     <div class="showcase-description-full">
-                        <h3>Artistic Style Transfer</h3>
-                        <p>Qwen-Image Edit excels at style transfer with support for a wide range of artistic transformations. From converting photos to impressionist paintings, applying anime aesthetics to real-world scenes, or transforming images to minimalist designs, the model adapts fluidly to creative editing requirements, making it a versatile tool for artists, designers, and storytellers.</p>
+                        <h3>Novel View Synthesis</h3>
+                        <p>Qwen-Image Edit excels at novel view synthesis, a key application in semantic editing. The model can rotate objects by various angles, including 90-degree and even full 180-degree rotations, allowing users to see different sides of objects. This capability is particularly valuable for product visualization, architectural rendering, and creative content production where multiple perspectives are needed.</p>
                     </div>
                     <div class="showcase-image-full">
-                        <img src="https://qianwen-res.oss-cn-beijing.aliyuncs.com/Qwen-Image/s2.jpg" alt="Artistic Styles Example" class="showcase-img-full">
+                        <img src="https://qianwen-res.oss-cn-beijing.aliyuncs.com/Qwen-Image/s2.jpg" alt="Novel View Synthesis Example" class="showcase-img-full">
                     </div>
                 </div>
 
                 <div class="showcase-item-full">
                     <div class="showcase-description-full">
-                        <h3>Advanced Image Manipulation</h3>
-                        <p>As a specialized image editing model, Qwen-Image Edit goes far beyond simple adjustments. It enables advanced operations such as comprehensive style transfer, object insertion or removal, detail enhancement, background replacement, and even human pose manipulation—all with intuitive input and coherent output. This level of control brings professional-grade image transformation within reach of everyday users.</p>
+                        <h3>Appearance Editing</h3>
+                        <p>Appearance editing is another powerful capability of Qwen-Image Edit. The model can keep certain regions of an image completely unchanged while adding, removing, or modifying specific elements. For example, it can insert signboards into scenes with corresponding reflections, remove fine details like hair strands, change the color of specific elements, or adjust backgrounds and clothing in portraits—all with exceptional attention to detail and realism.</p>
                     </div>
                     <div class="showcase-image-full">
-                        <img src="https://qianwen-res.oss-cn-beijing.aliyuncs.com/Qwen-Image/s3.jpg" alt="Image Editing Example" class="showcase-img-full">
+                        <img src="https://qianwen-res.oss-cn-beijing.aliyuncs.com/Qwen-Image/s3.jpg" alt="Appearance Editing Example" class="showcase-img-full">
                     </div>
                 </div>
 
                 <div class="showcase-item-full">
                     <div class="showcase-description-full">
-                        <h3>Intelligent Image Analysis</h3>
-                        <p>Qwen-Image Edit doesn't just transform—it understands. It analyzes the input image through a suite of understanding tasks, including object detection, semantic segmentation, depth and edge (Canny) estimation, and structural analysis. These capabilities ensure that transformations respect the original image's key elements and structure, resulting in edits that feel natural and coherent, powered by deep visual comprehension.</p>
+                        <h3>Text Editing Excellence</h3>
+                        <p>A standout feature of Qwen-Image Edit is its accurate text editing capability, which stems from Qwen-Image's deep expertise in text rendering. The model excels at editing both English and Chinese text in images, enabling modifications to large headline text as well as precise adjustments to small and intricate text elements. This makes it particularly valuable for poster design, advertisement creation, and multilingual content production.</p>
+                    </div>
+                    <div class="showcase-image-full">
+                        <img src="https://qianwen-res.oss-cn-beijing.aliyuncs.com/Qwen-Image/s4.jpg" alt="Text Editing Example" class="showcase-img-full">
+                    </div>
+                </div>
+                
+                <div class="showcase-item-full">
+                    <div class="showcase-description-full">
+                        <h3>Progressive Editing</h3>
+                        <p>Qwen-Image Edit supports chained, step-by-step editing approaches that allow users to progressively refine and correct images. For example, when editing complex calligraphy artwork, users can draw bounding boxes to mark specific regions that need correction and instruct the model to fix these areas one by one. This iterative approach enables precise control over the editing process, ensuring the desired final result is achieved even for challenging edits.</p>
                     </div>
                     <div class="showcase-image-full">
-                        <img src="https://qianwen-res.oss-cn-beijing.aliyuncs.com/Qwen-Image/s4.jpg" alt="Image Understanding Example" class="showcase-img-full">
+                        <img src="https://qianwen-res.oss-cn-beijing.aliyuncs.com/Qwen-Image/bench.png" alt="Progressive Editing Example" class="showcase-img-full">
                     </div>
                 </div>
 
@@ -222,12 +210,12 @@ image.save("example.png")</code></pre>
             <div class="resource-links-section">
                 <h2>Resources</h2>
                 <div class="resource-links">
-                    <a href="https://huggingface.co/Qwen/Qwen-Image-Edit" target="_blank" rel="noopener noreferrer" class="resource-link">Qwen-Image Edit on Hugging Face</a>
-                    <a href="https://github.com/QwenLM/Qwen" target="_blank" rel="noopener noreferrer" class="resource-link">Qwen GitHub</a>
-                    <a href="https://www.alibabacloud.com/en/solutions/generative-ai/qwen" target="_blank" rel="noopener noreferrer" class="resource-link">Alibaba Cloud Qwen</a>
-                    <a href="https://modelscope.cn/models/qwen/Qwen-Image" target="_blank" rel="noopener noreferrer" class="resource-link">ModelScope</a>
-                    <a href="https://help.aliyun.com/zh/dashscope/developer-reference/qwen-vlo-quick-start" target="_blank" rel="noopener noreferrer" class="resource-link">Qwen VLo Documentation</a>
-                    <a href="https://www.alibabacloud.com/help/en/model-studio/vision/" target="_blank" rel="noopener noreferrer" class="resource-link">Qwen-VL Documentation</a>
+                    <a href="https://wavespeed.ai/models/wavespeed-ai/qwen-image/edit" target="_blank" rel="noopener noreferrer" class="resource-link">Qwen-Image Edit on WaveSpeed</a>
+                    <a href="https://wavespeed.ai/models/wavespeed-ai/qwen-image" target="_blank" rel="noopener noreferrer" class="resource-link">Qwen-Image on WaveSpeed</a>
+                    <a href="https://wavespeed.ai/" target="_blank" rel="noopener noreferrer" class="resource-link">WaveSpeed AI</a>
+                    <a href="https://huggingface.co/Qwen/Qwen-Image-Edit" target="_blank" rel="noopener noreferrer" class="resource-link">Hugging Face</a>
+                    <a href="https://modelscope.cn/models/qwen/Qwen-Image-Edit" target="_blank" rel="noopener noreferrer" class="resource-link">ModelScope</a>
+                    <a href="https://chat.qwen.ai/" target="_blank" rel="noopener noreferrer" class="resource-link">Qwen Chat</a>
                 </div>
             </div>
         </div>