Z Image photoreal draft model
Use it when the image needs to read like a photo first: portraits, products, architecture, food, macro nature, and travel scenes.
Your generated image will appear here.
Enter a prompt and click 'Generate Image' — or start from an example:
Examples
Open any image for a closer look, then adapt the prompt to your own subject, composition, and style.
Photorealistic portrait
Rainy-window cinematic portrait
Prompt: Cinematic close portrait of an adult man in his early 30s seated beside a rain-covered cafe window at blue hour, wearing a charcoal wool coat, looking past the camera with a thoughtful expression. Warm pendant lights reflect as soft amber circles in the glass, cool rainy street outside, realistic pores and damp hair, 85mm lens, shallow depth of field, natural mixed lighting, restrained film color.
Automotive campaign
Electric motorcycle at dawn
Prompt: Photorealistic commercial photograph of an unbranded matte silver electric motorcycle parked on an empty concrete overlook at dawn. Low three-quarter camera angle, sharp wheel and brake detail, pale orange sky reflected in the body panels, distant city softened by morning haze, one cool edge light and a long natural shadow, premium automotive campaign composition, no rider, no logo, no text.
Macro nature image
Dew on fern macro
Prompt: Extreme macro photograph of a single curled fern tip covered in tiny morning dew droplets, one large droplet refracting an upside-down green forest, soft backlight creating a delicate rim along the leaf, shallow depth of field, realistic plant fibers and water surface tension, deep green background bokeh, nature-magazine detail.
Architecture visualization
Concrete courtyard home
Prompt: Photorealistic architectural photograph of a quiet modern courtyard house built from pale board-formed concrete and warm oak, a shallow reflecting pool running beside the glass living room, one olive tree casting patterned shadows on the wall. Early evening light, interior lamps just beginning to glow, straight verticals, 24mm lens, realistic materials and reflections, architecture magazine quality.
Restaurant editorial
Handmade dumplings
Prompt: Close food photograph of handmade vegetable dumplings in a dark ceramic bowl, one dumpling lifted with wooden chopsticks to reveal the pleated wrapper, chili oil shining in the folds, scallions and sesame scattered naturally. Warm side light, wisps of steam, realistic translucent dough texture, dark restaurant background, shallow focus, appetizing editorial styling.
Travel visual
Alpine train journey
Prompt: Photorealistic wide travel photograph from a hillside above a red narrow-gauge train curving across a stone viaduct in the Swiss Alps. Fresh snow on high peaks, deep green valley below, late-afternoon sun breaking through clouds in visible shafts, realistic scale and atmospheric perspective, crisp train detail without toy-like proportions, panoramic editorial composition.
Model context
Z Image Turbo is a speed-oriented photoreal image model. Its useful range is not abstract graphic design, but believable light, materials, people, interiors, and places.
On RenderFlow AI, it sits between low-cost drafting and premium finishing: quick enough for exploration, but strong enough to judge whether a photographic direction works.
Model strengths
Use it when the image needs to read like a photo first: portraits, products, architecture, food, macro nature, and travel scenes.
Prompt guide
Name the lens feel, distance, angle, and depth of field so the result lands as photography rather than generic art.
Say where light comes from and whether it is soft, hard, mixed, dawn, blue hour, or studio.
For products and vehicles, request no logos or text unless you plan to proofread and regenerate.
Reject outputs with wrong hands, toy-like vehicles, or impossible reflections before judging mood.
It can illustrate, but its clearest value is fast photoreal generation for commercial and editorial scenes.
A good prompt reads like a photo brief: subject, location, camera, lighting, material details, and a clear exclusion for logos or unwanted text.
Yes, especially unbranded product concepts with simple surfaces, controlled backgrounds, and realistic material notes.
Use GPT Image 2 or Qwen Image for text-heavy layouts, and Seedream 5.0 or Imagen-style models when final polish is more important than speed.