Z-Image_clear_anime Released! What Will the Next Generation of Anime Models Look Like?

Introduction

Hello, I'm Easygoing.

This time, I'm introducing the Z-Image_clear_anime model, a high-speed anime model that's completely free to use under an open license.

Z-Image (Base) + Z-Image-Turbo!

Z-Image_clear_anime is a custom model created by merging Z-Image (Base) and Z-Image-Turbo specifically for anime illustrations.

flowchart LR

subgraph Alibaba Group

A1("Z-Image (Base)")
A2("Z-Image-Turbo")

end

subgraph Easygoing Models

B1(Z-Image-Turbo_clear)
C1(Z-Image_clear_photoreal)
C2(Z-Image_clear_anime)

end

A1-.->A2
A2-.->B1
A1-.->C1
B1-.->C1
A1-.->C2
B1-.->C2

• Z-Image (Base): Focuses on image quality and variety
• Z-Image-Turbo: Emphasizes fast generation and stability

Last time, I introduced the Z-Image_clear_photoreal model, which merged Z-Image for photorealistic purposes.

Since Z-Image requires completely different adjustments for photorealistic and anime styles, the Z-Image_clear_anime model I'm introducing this time has been re-tuned specifically for anime illustrations.

Actual Illustrations!

Now, let's take a look at some actual illustrations.

On the left is the newly tuned Z-Image_clear_anime, and on the right is the original Z-Image-Turbo_clear.

Perfume Bottle

Dusk

Close-Up

The left side, Z-Image_clear_anime, features clearer color expression compared to the right side's Z-Image-Turbo_clear, and when looking at the overall illustration, you can see that the sense of depth is better organized.

Recommended VAE: Z-Image_natural_vae

For Z-Image_clear_anime, I recommend using the custom model's Z-Image_natural_vae for better color expression.

Z-Image_natural_vae is a model adjusted to increase saturation while maintaining natural color expression compared to the original Flux_ae.safetensors.

Since Z-Image's anime illustrations tend to have higher saturation than photorealistic ones, among the VAEs, Z-Image_natural_vae strikes just the right balance.

What Will the Next Generation of Anime Models Be Like?

Anime illustrations are not limited to Japan or East Asia; they are now one of the most popular genres worldwide.

Currently, the most popular anime models for local image generation are custom models based on Stable Diffusion XL.

gantt
    title Next-Gen Anime Model Comparison
    dateFormat YYYY-MM-DD
    axisFormat %Y
    tickInterval 12month

	Stable Diffusion XL : done, 2023-07-26, 2026-02-16
	Lumina-Image-2.0 : done, 2025-01-25, 2026-02-16
	Z-Image : 2025-11-25, 2026-02-16
	Flux.2 [klein] : done, 2026-01-15, 2026-02-16
	Anima : done, 2026-01-29, 2026-02-16

• Lumina-Image-2.0: From the Neta-Lumina project
• Z-Image: Good balance of size and quality
• Flux.2 [klein]: Latest generation VAE
• Anima: Backed by ComfyUI development

There are several image generation AIs expected to succeed SDXL for anime illustrations, but Z-Image stands out as a top candidate due to its good balance of size and quality, along with its open license.

Future performance improvements will likely depend on the release of large-scale fine-tuned models comparable to Animagine-XL or Illustrious-XL.

With the release of Z-Image (Base), optimized for training, in January 2026, I'm looking forward to what's next.

Summary: Looking Forward to the Future Development of Z-Image

• Z-Image_clear_anime is dedicated to anime
• Improved depth and color expression
• Excited for future developments

This time, I introduced the Z-Image_clear_anime model. Z-Image offers a great balance of size and image quality, with a free license, making it a model with high expectations for future growth.

Anime illustrations, starting from Japan and East Asia, are now one of the most anticipated genres in image generation worldwide, so I'll continue to follow the latest updates.

Thank you for reading to the end!