Prompt: make the subjects skin details more prominent and natural make the subjects skin details more prominent and natural

To use this LoRA, you need to load the base model Qwen/Qwen-Image-Edit-2509, and then apply the fine-tuned LoRA weights loaded as qwen-edit-skin.safetensors. Previous versions of the weights have been uploaded for reference, but the final version is qwen-edit-skin.safetensors. You can also utilize the example ComfyUI workflow attached in the repository to compare results across different weights.

The recommended weight is between 1 and 1.5. The provided examples showing weights up to 2 are solely to demonstrate the effect of the LoRA, as a strength that high is considered too intense for practical use.

A fine-tuned LoRA for Qwen-Image-Edit-2509 to enhance skin realism. You can find a more in-depth article on LinkedIn detailing my approach to this LoRA: This repository contains a fine-tuned Low-Rank Adaptation (LoRA) model designed to enhance the realism and detail of human skin in images. Trained on the powerful Qwen/Qwen-Image-Edit-2509 base model, the LoRA leverages its advanced image editing capabilities to focus on generating more natural and detailed skin textures.

The model was trained for 5,000 steps on a local RTX 5090 GPU using AI Toolkit. The resulting LoRA model is ideal for photographers, digital artists, and anyone looking to elevate the quality of human subjects in their generated or edited images.

Model Description The qwen-edit-skin LoRA is a specialized fine-tuned version of the Qwen/Qwen-Image-Edit-2509 base model. The base model is a powerful image editor that excels at multi-image editing and maintaining consistency within a single image, particularly in preserving individual identity details. Building upon this, this LoRA specifically optimizes for the nuances of human skin, adding details and realism that the original version might lack.

Training used a forked version of AI ToolKit, a comprehensive toolkit for fine-tuning diffusion models. The dataset curation process involved reverse-modifying the subject's skin details through the following steps:

• Took real portrait photographs of various subjects, including exposed skin.
• Labeled these images as our "target" (final result), representing the expected final outcome in a standard Qwen Edit workflow.
• Edited the images in Photoshop by applying additional Gaussian blur and softening skin tones, making skin textures, tones, and pores less prominent.
• These became our "control" (before) images used for Qwen Edit training.