Image

The image scenarios are divided into three types: Text-to-Image, Image-to-Image, and Inpainting.

1. Text-to-Image

In the Lab, select the Text-to-Image scenario from the left sidebar. The center area will display the Text-to-Image section. In the dropdown menu, choose the graphic model you want to use. Enter your prompt for the image you want to generate in the prompt box. If there are multiple conditions/prompts, you can separate them with commas.

Below the prompt box is the Negative Prompt, which is used to specify elements or features you wish to avoid in the final image. Users can explicitly tell the model what should not appear in the image. This helps control the quality and details of the output image, preventing the generation of images with undesired features.

For example:

• Prompt: “Mountains under a bright blue sky”

• Negative Prompt: “No people, no houses, no clouds”

With this Negative Prompt, the system will attempt to avoid generating images that include people, buildings, or clouds, bringing the result closer to the user's expectations.

After entering the prompt and negative prompt, the generated image will appear in the central interaction area.

The right sidebar is the parameter settings area for Text-to-Image. Depending on the model series, different configurable parameters may be supported. Refer to the Glossary & Concpetion to adjust the parameters according to your needs for optimal generation results.

1. Image-to-Image

In the Experience Center, select the Image-to-Image scenario from the left sidebar. The interaction in Image-to-Image is similar to Text-to-Image, with the main difference being that below the Negative Prompt box, users can upload an original image. The model will then combine the uploaded image with the entered prompt and negative prompt to generate a new image on the right side.

1. Inpainting

In the Lab, select the Inpainting scenario from the left sidebar. Inpainting is an extension of the Image-to-Image interaction. With Inpainting, users can modify specific areas of an image while maintaining the overall naturalness and continuity of the image. Below the Negative Prompt box, users can upload an original image, as well as a mask image. If no mask image is provided, the interface offers two options:

• Automatically Generate a Mask Image (Users can click the appropriate button based on whether they want to modify the foreground or the background).

• Generate a Mask Image Using a Hand-Drawn Brush: (i) Use the brush tool to paint and outline the mask. （ii) Use the eraser tool to remove any incorrect areas to make the mask more precise.

Finally, the model will combine the uploaded original image, mask image, and the entered prompt and negative prompt to generate a new image on the right side.