Which parameters are supported?

This page supports prompt, width, height, Thinking Mode, and seed. Width and height are submitted as a combined size value.

What does Thinking Mode do?

It gives the model more reasoning time, which usually improves complex scenes and detail quality, but it also increases generation time.

The current form supports width and height from 512 to 4096, with a 64-pixel step and a default size of 2048 by 2048.

A seed of -1 means random generation. Use a fixed integer if you want reproducible results.

A practical text-to-image workflow for high resolution, prompt reasoning, and repeatable results.

Give the model more reasoning time for complex materials, lighting, composition, and atmosphere.

Use width and height sliders to build large square, vertical, horizontal, or poster-style outputs.

Keep a direction stable while refining prompts instead of restarting from a new random result every time.

Works well for portraits, products, concept scenes, fashion visuals, and other detail-heavy ideas.

State the subject, shot type, and composition first so the model locks onto the right structure.

Precise words for surface, texture, and light direction usually improve fidelity.

It is especially helpful when you generate at 2048 or above or need lots of fine detail.

Use random generations to search, then switch to a fixed seed for controlled iteration.

Create close-up portraits, editorial concepts, and beauty campaign directions.

Generate product heroes, key visuals, covers, and marketing images.

Test worldbuilding, color language, material studies, and cinematic moods quickly.

Adapt outputs to square, vertical, or wide layouts for banners, covers, and posters.

Common questions about Wan 2.7 text to image.

Open the Wan 2.7 creator form and test how custom resolution and Thinking Mode change your outputs.