p.kaczmarek2 wrote:By the way, do you know about conditional prompting? If I remember correctly, it was the keyword BREAK in the prompt.
It depends what you mean by this:
1. Keywords in the prompt text - I don't know anything about that.
2. Prompt engineering, i.e. crafting the text with standard control - I know about the possibility of weighting factors for Stable Diffusion and I know that this is a simple generator that can't cope with complex wording, but can partly be dealt with by specially adjusting the wording in terms of vocabulary and syntax. Then, what words to add so that the expected image style comes out.
3. ControlNet, masks, combining several prompts, attaching an image - this can be done in ComfyUI, I know how it works and how to do it, but it's more of an advanced and conditional control of the model action, not prompt engineering.
