Skin Enhancer

Skin Enhancer is best for restoring authentic skin pores, fine lines, natural blemishes, and subtle wrinkles onto AI portraits that look waxy or airbrushed. It maps high-frequency micro-textures directly onto facial lighting and contours without altering underlying facial geometry.

Adjust Image Strength toward higher values (0.7 to 0.9) to lock down the original facial likeness, while setting LoRA Strength between 0.8 and 1.2 for balanced skin grain. If your input is heavily smoothed, raise LoRA Strength up to 2.0; lower Image Strength below 0.4 only when you want the text prompt to actively alter facial structure.

Avoid using Skin Enhancer on cartoon illustrations, non-human characters, or heavily stylized graphic art, as the model aggressively maps human pores onto non-photorealistic surfaces. It should also be avoided for text rendering or complex background alterations, as the distilled inference pipeline specializes strictly in facial detailing.

Choose PNG for maximum preservation of fine micro-pores and delicate wrinkles when preparing assets for print or high-end retouching workflows. Select JPEG or WebP when you need compact file sizes for fast web delivery or interactive previewing.

Skin Enhancer is developed by Alibaba Group (Tongyi-MAI) and built on Z-Image Turbo, a 6-billion parameter Scalable Single-Stream Diffusion Transformer (S3-DiT) paired with a specialized skin-detailer LoRA. Released in late 2025, it delivers studio-grade micro-detailing in sub-second inference passes.