Text-to-Image Guide

Text-to-image generation creates an image from a written description without a reference image. Describe the subject, setting and visual style, then select a model and output settings to generate the image.

1. Select a model

Open the workspace and select an available image model. For text-to-image generation, leave the Reference images area empty. If an image is already attached, remove it before submitting.

For a comparison of reference-image support, editing and output controls, see the Model Selection Guide.

2. Write the generation prompt

Describe the main subject and its key features, followed by the setting, lighting, composition and visual style. Include any required empty space or elements to exclude.

Product photography example: “A tan leather handbag with gold hardware on a cream stone table beside a window. Soft natural side lighting, realistic leather texture and a simple background. Square composition with the bag centered. No text or watermark.”

Illustration example: “An editorial illustration of a reading corner with a wooden bookshelf, a green armchair and a floor lamp. Warm muted colors, clean shapes and subtle paper texture. Portrait composition with clear space at the top for a heading.”

Use specific visual descriptions. When several requirements conflict, prioritize the main subject and composition before adding secondary details.

3. Configure output settings

Select an aspect ratio for the intended use: square for a product tile, portrait for a poster or landscape for a banner. Available resolution, quality and image-count options depend on the selected model.

Review the Credit cost shown on the Generate button after changing the settings, then submit the task.

4. Review and refine the result

Check the subject, composition, text and local details before downloading. If the result differs from the intended direction, revise the prompt and generate another image.

For a focused adjustment to an existing result, select Continue editing on a supported image. To guide generation with an existing photo, follow the Image-to-Image Guide.

Troubleshooting

Missing subject details: place the main subject and its required features near the beginning of the prompt, and reduce unrelated descriptions.

Inconsistent style: use one clear visual direction and remove conflicting style requirements.

Task unavailable or failed: check model availability, your Credit balance and the displayed error. Wait for the current task to finish before submitting another request.