ImageByQwen.com – Fast AI text-to-photo using Qwen It utilizes the Qwen-Image-Layered model (latest open-source release by the Alibaba Qwen team in December 2025) to intelligently decompose any single image (JPG/PNG/WebP) into 3-8 transparent RGBA layers, each with a perfect alpha channel, soft edges, and correct occlusion order. Key highlights: Automatic semantic segmentation: Intelligently identifies foreground, background, objects, text, shadows, etc., automatically filling in occluded areas without manual cutout.

Controllable decomposition: Specifies the number of layers (3/4/5/8), supporting recursive subdivision (unlimited levels).

Editing freedom: After importing into Photoshop/Figma/After Effects, objects can be independently moved, scaled, recolored, and deleted, maintaining pixel consistency.

Professional-grade output: Transparent PNGs, perfectly preserving soft edges such as hair, glass, and smoke.

API-friendly: No cold start, batch processing, suitable for e-commerce, design, VFX, animation, and other scenarios.

Real-world example: Upload product photos → Break them down into 4 layers: product, background, text, and lighting → Easily change the background or color in seconds.