This template is a compact, text-to-image workflow built around the Z-Image Turbo generator, designed to get from prompt to output with minimal setup. At its core is a single custom Z-Image Turbo T2I node (node id f2fdebf6-dfaf-43b6-9eb2-7f70613cfdc1) that encapsulates text encoding and image generation, paired with a SaveImage node to write results to disk and a MarkdownNote for in-graph guidance. The template is configured to use the Qwen 3 4B text encoder weights (qwen_3_4b.safetensors) for prompt conditioning.

Technically, the Z-Image Turbo node handles the end-to-end generation: it ingests your prompt (and optional negative prompt), encodes it with the specified text encoder, and runs the model’s internal sampler to produce an image tensor. That tensor is passed directly to SaveImage, which exports PNGs to your ComfyUI output folder using the configured filename prefix. Because the heavy lifting is wrapped in a single node, you get a clean, reliable T2I pipeline that’s easy to iterate: tweak prompts, resolution, seed, and sampling controls in one place, then save consistent, reproducible outputs.