Skip to main content
This guide walks through the complete image generation process in ZeroTwo, including step-by-step instructions, prompt writing tips, and how to work with your generated images.

​
Step-by-step: generate an image

1

Navigate to the image workspace

Click Images in the topbar or go to /studio/images. You’ll see the prompt field, controls, and your generation gallery.
2

Write your image description

Enter a description of the image you want in the prompt field. Be specific: include the subject, style, lighting, mood, and composition.Example prompts:
  • A serene Japanese garden at sunset with cherry blossoms and a stone lantern, soft warm lighting, photorealistic
  • Minimalist product shot of a white ceramic coffee mug on a marble surface, studio lighting, top-down view
  • Fantasy landscape with floating islands and waterfalls, digital art, vibrant colors, wide establishing shot
3

Select a model

Open the model dropdown and choose an image model. Different models have different strengths — see image models for guidance. GPT-Image-1.5 and Imagen 4.0 are strong defaults for photorealism; Flux Pro v1.1 excels at artistic styles.
4

Choose a style preset (optional)

Select a style preset from the style selector. Presets add visual guidance to your prompt — for example, choosing “Watercolor” biases the output toward that aesthetic without requiring you to describe it in detail. See styles and settings for all available presets.
5

Set the aspect ratio

Choose the aspect ratio that matches your intended use:
  • Square (1:1) — social media posts, profile images
  • Landscape (16:9) — blog headers, presentations, desktop wallpapers
  • Portrait (9:16) — mobile wallpapers, Instagram Stories, vertical content
  • Other ratios — available depending on the selected model
6

Set the number of images

Choose how many images to generate — typically 1 to 4 per prompt. Generating 4 variations with the same prompt is a great way to explore different interpretations and pick the best result.
7

Click Generate

Click the Generate button. Generation usually takes a few seconds to about 30 seconds, depending on the model, quality setting, and number of images. A progress indicator shows while the images are being created.
8

Review your images in the gallery

Generated images appear in the gallery below the controls. Hover over any image to reveal action buttons.

​
Prompt tips for better images

Writing a strong prompt makes a significant difference in output quality. Here are practical techniques:

​
Be descriptive about the subject

Instead of: a dog Try: A golden retriever puppy sitting in a field of sunflowers, looking up at the camera, soft afternoon light

​
Specify the style explicitly

  • "in the style of watercolor illustration, soft edges, pastel tones"
  • "ultra-realistic photograph, 85mm portrait lens, shallow depth of field"
  • "Studio Ghibli anime style, lush green scenery, gentle light"
  • "oil painting, textured brush strokes, impressionist"

​
Describe the composition

  • "close-up portrait, face fills the frame"
  • "wide establishing shot, subject small in the scene"
  • "bird's-eye view looking straight down"
  • "low angle shot looking up at the subject"
  • "rule of thirds, subject slightly off-center"

​
Mention the lighting

Lighting has a major impact on the mood and realism of the image:
  • "golden hour light, warm orange tones"
  • "dramatic studio lighting with deep shadows"
  • "soft diffused natural light from a north-facing window"
  • "neon city lights at night, colorful reflections"
  • "overcast flat light, even shadows"

​
Combine elements for best results

A strong prompt combines subject + style + composition + lighting: A weathered lighthouse on a rocky coastline at dusk, waves crashing below, dramatic storm clouds, long-exposure photograph style, cool blue-grey tones, wide establishing shot
Generate 4 variations of the same prompt to see different interpretations. Even with identical text, models produce meaningfully different results — generating multiples helps you find the best one quickly.

​
Gallery actions

Once images are generated, hover over any image in the gallery to reveal these actions: Your generation history is persistent — previously generated images remain in the gallery across sessions so you can revisit and download them later.

​
Generating variations

To explore variations of a specific result:
  1. Note the prompt used for the image you like
  2. Click into the prompt field and make minor adjustments (or keep it the same)
  3. Generate again — even the same prompt produces different outputs
Some models offer a “regenerate” or “variations” option directly from the gallery image. Check the model-specific options when hovering over a generated image.
Prompt changes don’t have to be large. Swapping one word (e.g., “sunny” to “overcast”) or adding a single detail (“with a red umbrella”) can significantly change the output direction.