Announcing Qwen-Image-Edit-2511_clear! The Latest Model for Vivid and Sharp Color Expression

Introduction

Hello everyone, I’m Easygoing.

Today, I’m excited to introduce my latest release: Qwen-Image-Edit-2511_clear, a model fine-tuned for vivid colors and sharp, high-contrast visuals.

What is Qwen-Image-Edit-2511?

The Qwen-Image-Edit series is a line of AI image editing models released by the Alibaba Group in China.

gantt
    title Image Editing AIs
    dateFormat YYYY-MM-DD
    axisFormat %Y-%m
	tickInterval 6month

    section Open AI
        GPT-Image-1 : done, 2025-04-23, 2026-01-09
        GPT-Image-1.5 : done, 2025-12-17, 2026-01-09

    section Black Forests Labs
        Flux.1-Kontext: done, 2025-06-26, 2026-01-09
        Flux.2: done, 2025-11-25, 2026-01-09

    section Google
        nano-banana : done, 2025-08-26, 2026-01-09
        nano-banana-pro : done, 2025-11-20, 2026-01-09

    section Alibaba
        Qwen-Image-Edit :  done, 2025-08-17, 2026-01-09
        Qwen-Image-Edit-2509 :  done, 2025-09-22, 2026-01-09
        Qwen-Image-Edit-2511 :  2025-12-17, 2026-01-09

Among them, Qwen-Image-Edit-2511 is the most recent iteration, officially released on December 17, 2025.

Major Improvements in Qwen-Image-Edit-2511

Compared to its predecessor, the 2509 version, Qwen-Image-Edit-2511 features comprehensive enhancements across the board.

Key Improvements in Qwen-Image-Edit-2511:

• Overall boost in image quality
• Better character consistency
• Integration of popular community-made LoRAs

Qwen-Image-Edit-2511 represents a genuine evolution in overall quality.

Character consistency has also seen significant improvements. Faces and outfits remain more stable during the editing process, making it much more practical for iterative design work.

Furthermore, by leveraging the benefits of its open license, the model now integrates popular community-developed LoRAs, greatly improving usability in specific scenarios.

Popular LoRAs Integrated into Qwen-Image-Edit-2511

Lighting Adjustments (Relight)

Perspective/Camera Angles

Comparing Qwen-Image-Edit-2511 Visuals

Let’s take a look at how the model actually performs.

In the four-way comparisons below, the top row is the 2511 model, and the bottom row is the 2509 model. Additionally, the left side shows the "clear" model (my custom-tuned version), while the right side shows the original Qwen-Image model.

Sunset in the City

Sunset in the Mountains

Qwen-Image-Edit-2511: An All-Around Evolution

Comparing the right-side images (2511 vs. 2509), you can see that 2511 produces much clearer results with significantly less noise—a common artifact in many diffusion models.

The Qwen-Image-Edit-2511 model also demonstrates better prompt adherence, making it a more reliable and polished tool for AI image generation.

The "clear" Model: Sharp and Vivid

Now, let’s look at the "clear" models on the left.

With the previous 2509_clear model, I made extensive adjustments to improve the overall output, which resulted in a look quite different from the original.

For the 2511 version, however, the original was already highly polished. Therefore, I focused my tuning primarily on color expression and contrast. As a result, 2511_clear delivers an exceptionally sharp and vibrant output.

More Sample Works

Here are some additional examples generated using the Qwen-Image-Edit-2511_clear model.

Night Street

 realistic, photorealistic,

 a female with blonde hair stands on a wet street at night surrounded by city lights with a car parked nearby bathed in the soft glow of streetlights and the glow of streetlights and reflections on the wet road,

dynamic angle, dutch angle, upper body, close up, little smile, laugh, peaceful, wind, romantic, alizarin, burgundy, rouge, maroon, indigo, royal blue, deep blue, deep purple, royal purple

Late Night Streaming

 anime, japanese anime,

 a blond female sits at a desk with a laptop, microphone, wearing a purple dress and headphones, in a dimly lit room with a nighttime cityscape visible through the window, with a starry sky visible through the window.,

 dutch angle, dynamic angle, upper body, close up, happy, peaceful, colorful, vivid, burgundy, rouge, alizarin, royal purple, indigo, deep blue, black background, simple background,

Afternoon Tea

realistic, photorealistic,

a female sits in a cozy room with a city view, surrounded by books, a lamp, and flowers, wearing a blue and white striped blouse with a floral pattern and a blue skirt with a floral pattern, with a contemplative expression on her face and a cup of tea in front of her.,

dutch angle, dynamic angle, upper body, close up, happy, peaceful, colorful, vivid, burgundy, rouge, alizarin, royal purple, indigo, deep blue, black background, simple background,

VAE: Qwen-Image_clear_vae

All the samples shown here were output using Qwen-Image_clear_vae, an improved version of the original VAE.

Compared to the original, Qwen-Image_clear_vae is adjusted to produce healthier skin tones by enhancing the red spectrum. If you’re trying out the Qwen-Image-Edit-2511_clear model, I highly recommend using this VAE alongside it.

Summary: Give Qwen-Image-Edit-2511_clear a Try!

• Qwen-Image-Edit-2511 is a significant, "proper" evolution.
• Qwen-Image-Edit-2511_clear brings enhanced color expression and clarity.
• The model continues to improve thanks to community feedback.

In this post, we explored the features of Qwen-Image-Edit-2511_clear.

Thanks to its open license, Qwen-Image-Edit-2511 has benefited from community feedback to become a top contender in the AI image editing space.

While Qwen-Image is known for being resource-intensive, Alibaba is also developing the lightweight Z-Image-Edit model. If that also arrives with an open license, Alibaba will likely continue to lead the initiative in AI image editing.

I hope you take this opportunity to experiment with the newly refined Qwen-Image-Edit models!

Thank you for reading until the end!