As a general rule, when using a custom generator that you want to keep stylistically similar to your training images, it is best to keep your prompts limited.
The training process provides more context for the program to draw from, which reduces the need for extra prompts.
When prompting a generator, it's important to provide a clear and specific description of the image you want it to create. This may include information such as the subject or objects to include in the image, the overall composition and layout, and any relevant details or constraints.
It is also important to provide the model with the necessary context or background information to help it generate an accurate and realistic image. This may include information about the environment, lighting, or other factors that affect the appearance of the image.
To make the prompt easy for the generator to understand and interpret, it is important to use clear and consistent language and avoid ambiguity or confusion.
The goal of prompting a generator for image generation is to provide it with the information and context it needs to create an accurate image that meets the requirements of the task.
If you have designed a character or subject that has not been trained in a specific default style, you may choose to use more prompts.
You can also find various prompt guides through the following links:
Prompthero
Lexica
Stack Diary
Stable Diffusion Art
Stable Cog
Jon Stokes
CLIP Interrogator