We train the first model on created images or your dataset, then test it. It will respond to minimal prompts (even single words) to produce results as close as possible to the original dataset (input graphics used for training).
This model is usually sufficient for non-designers.
The second model is adapted for more flexible requests depending on the tasks, such as creating different styles (levels of realism, vector graphics, etc.). This model suits advanced users, including designers.
Both models can be combined. We will also provide detailed training materials specific to your case.
This set is sufficient to cover all your needs. If you realize that additional customization (e.g., a third model) is required, we can provide it as well.