This workflow is based on the official Qwen Image workflow and has been modified to include multiple LoRAs to enhance image quality. The LoRAs can be freely selected, but it is not recommended to use more than 4 LoRAs simultaneously. It also includes an image-to-image section, a ControlNet section, and three high-definition upscaling sections (Ultimate SD Upscale, SeedVR2 Upscale, and the traditional upscaling using an enlargement model, labeled as "CR Upscale" in the workflow). Users can choose according to their actual needs. Note that the number of LoRAs affects adherence to prompts—the more LoRAs used, the less the image results may adhere to the prompts, which requires users to balance accordingly.

Upgraded V2 Workflow Linkhttps://www.runninghub.cn/post/1992255184711602177