Is Qwen-Image-Layered free?

Yes, it is an open-source model released under the Apache 2.0 license, meaning it is free for both research and commercial use.

How is this different from Segment Anything (SAM)?

SAM only creates "masks" (selections). Qwen-Image-Layered creates full RGBA images with transparency and fills in the background behind objects (inpainting), so you can actually move them around.

Can I use this in Photoshop?

While it doesn't run inside Photoshop natively yet, the layers it generates can be exported as PNGs and dropped directly into Photoshop or any other editing software.

What hardware do I need?

It is a heavy model. For optimal performance at 1024px resolution, high-end GPUs (like an RTX 4090 or A100) are recommended, though lower VRAM versions (FP8) are available for consumer cards.

Qwen Image Layered

Layered Decomposition for Inherent Editability.

From Alibaba Cloud

Founder: Qwen Team (Alibaba Cloud)

Image & Design

Dec 2025

63k+

Free

Pricing Model

Free

Starting Price

$0/Month

About This AI

Qwen-Image-Layered is a groundbreaking AI model that bridges the gap between flat AI images and professional design workflows.

Unlike traditional image generators that produce a single flattened JPEG, this model “decomposes” any image into multiple transparent (RGBA) layers—separating foreground, background, text, and objects.

This allows users to independently move, resize, recolor, or delete specific elements without affecting the rest of the image, effectively giving every AI-generated image “Photoshop-like” editability by default.

AI Art Image Editor LLM Text To Image

Pricing

Pricing Model

Free

Starting Price

$0/Month

Key Features

Layer Decomposition: Automatically splits a single image into 3–8+ distinct RGBA layers (e.g., separating a character from the background).
Occlusion Handling: intelligently "inpaints" the hidden areas behind objects so that moving a foreground element doesn't leave a hole.
Recursive Decomposition: Any isolated layer can be further broken down into sub-layers (e.g., splitting a "face" layer into eyes, nose, and mouth) for infinite precision.
High-Fidelity Editing: Supports non-destructive operations like resizing, repositioning, and recoloring individual layers.
Alpha Channel Generation: Creates clean, transparent cutouts for every element, ready for compositing.
Text Isolation: Can separate text from the background, allowing for easy translation or font changes without ruining the artwork.

Pros

Completely free and open-source (Apache 2.0).
Solves the "flat image" problem of Generative AI.
"Matryoshka-style" infinite editing depth.
Compatible with professional workflows (can export to PSD/compositing tools).
High consistency when editing specific parts of an image.

Cons

High hardware requirements (Requires significant VRAM, ~45GB for top performance, though FP8 versions exist).
Generation speed is slower than standard "flat" models.
Setup requires technical knowledge (Python/ComfyUI) rather than a simple one-click app.

Best for Graphic designers, game developers, and professional editors who need AI assets that are actually editable and compositable, rather than just static images.

FAQs

Is Qwen-Image-Layered free?
Yes, it is an open-source model released under the Apache 2.0 license, meaning it is free for both research and commercial use.
How is this different from Segment Anything (SAM)?
SAM only creates “masks” (selections). Qwen-Image-Layered creates full RGBA images with transparency and fills in the background behind objects (inpainting), so you can actually move them around.
Can I use this in Photoshop?
While it doesn’t run inside Photoshop natively yet, the layers it generates can be exported as PNGs and dropped directly into Photoshop or any other editing software.
What hardware do I need?
It is a heavy model. For optimal performance at 1024px resolution, high-end GPUs (like an RTX 4090 or A100) are recommended, though lower VRAM versions (FP8) are available for consumer cards.