Image Prompt Rules
Although different image generation applications may handle prompts differently, the basic rules are quite similar:
- Word Order: The order of words in the prompt determines their weight. Words that appear earlier in the prompt have a higher weight.
- Quantity Control: The number of objects should generally be kept between two and three to avoid overwhelming the model with too much information.
- Element Emphasis: To emphasize a specific element, use parentheses or exclamation marks. For example,
beautiful forest background, desert!!, (((sunset)))
will prioritize "desert" and "sunset" elements. Weighted values inside parentheses (e.g., (masterpiece:1.5)
) specify the importance of the details, and multiple parentheses have a similar effect.
- Fusion Expression: Use square brackets to indicate the blending of different elements. For example,
{blue hair:white hair:0.3}
represents a combination of blue and white hair, with blue hair having a 0.3 weight.
- Negative Prompts: Negative prompts effectively help avoid unwanted elements. For example, to generate an image of a mountain without trees, you can use
mountain | tree:-10
, where tree:-10
significantly reduces the weight of trees.