AI image portfolio / ComfyUI

An image is only part of the story.

Explore images made with Z-Image Turbo and the processes behind them. One workflow starts with a written prompt; the other extracts a prompt from a reference image before generating a new image. Both include facial refinement and two stages of upscaling.

Prompt → ImageImage → Prompt → ImageComfyUI

Black-and-white portrait of a model turning toward the camera against a dark background

SELECTED WORK / 01

Selected images

Selected works

Monochrome studies of form and light, followed by color portraits in natural light. Select an image for a larger view.

01 / Black & white

Portraits in light and shadow

Black-and-white figure standing beside a large opening in a concrete building — 01

Model seated beside a window in a sunlit room — 02

Model crossing a city street at night — 03

Close black-and-white portrait of a model in dark clothing — 04

Model seated on steps in strong light and shadow — 05

02 / Color portraits

Color and character

Side portrait of a model wearing a lilac sweater — 08

Model standing in warm window light — 09

Model in a light knit top and long skirt — 10

Model wearing a black outfit and hair ribbons — 11

03 / Landscapes & spaces

Landscapes and spaces

Studies of light, texture, and depth beyond portraiture.

Mountain valley and winding river in morning light — 12

City street at night after rain, with reflections from streetlights — 13

Cobblestone alley leading toward a bright opening — 16

Two workflows

Two ways to create

The starting point changes, while each ComfyUI workflow carries the image through its finishing stages.

Start with a prompt

Describe the subject, clothing, setting, and composition in text, then generate an image with Z-Image Turbo.

Text prompt → Z-Image Turbo → Generated image

Start with an image

Gemma4 describes the visual features of a reference image as a prompt. Z-Image Turbo then uses that prompt to generate a new image; this workflow does not directly edit the reference.

Reference image → gemma4 prompt extraction → Z-Image Turbo generation

Both workflows

Finishing stages

Face Detail refines facial features, while Face Swap replaces the face. Two upscaling processes complete the output.

Production setup

Production tools

WebUI: ComfyUIImage model: Z-Image TurboImage analysis: gemma4Face DetailFace SwapTwo-stage upscaling