This note describes a LoRA module designed for the image generator Qwen IE 2509 with a focus on changing materials and textures of scene elements.

Workflow and inputs

The module expects two uploaded images: the first provides the object or environment to be modified, while the second serves as a visual reference for target materials.

Users prepare prompts that instruct the model to transfer surface characteristics and color palettes from the reference image to the source image, preserving composition and lighting as much as possible.

Example prompt trigger

The recommended concise trigger for production prompts is provided below and is used verbatim to control the retexturization directive.

«change materials of image1 to match the reference in image2»

License and usage

The module is distributed under the Apache 2 commercial license, allowing integration into commercial pipelines subject to the license terms and attribution requirements.

Implementers should verify compatibility with their deployment stack and ensure that model weights and LoRA adapters are applied according to the image generator’s adapter workflow to avoid inconsistencies.

Notes on results and limitations

Quality depends on the similarity of viewpoint and lighting between the two images, as well as the level of prompt specificity. Complex materials may require iterative prompt refinement and additional masks.

Performance and artifacts vary by model version and hardware; testing on representative samples is recommended before large-scale application of the module.