ComfyUI – Z‑Image Workflow

This workflow is a customized extension of the default setup. It begins by feeding the input image into the Qwen‑VL node, which analyzes and generates a textual description of the image. From a predefined list, the workflow then selects a random style and combines it with the description. The resulting prompt is sent to the Z‑Image Turbo node, which synthesizes a new image that blends elements of the original with the chosen style.

I use QwenVL-4b in the node.