Prompt Length refers to the character count in each part of a multi-prompt. It provides the flexibility to control the detail of each part.

Multiprompt Parts, also known as "prompts", are individual segments of a multi-prompt. Each part is separated by double colons ::. This allows exploration of multiple concepts simultaneously.

Space Frequency dictates the insertion frequency of spaces in the generated text. Potential outcomes include:

• 0% Generation of long, complex words without spaces.

• 100% Production of multiple shorter words or single characters, each separated by a space.

Permutations denote the number of unique multi-prompt variations generated at once. This feature is useful for exploring a wide array of prompt combinations.

Alternating Weights assign negative weight to every other prompt in a multi-prompt, balancing the total weight around zero. This mechanism guides the model to generate varied results by indicating what to include and what to avoid.

When New Seed on Copy is activated, a new random seed is generated for every copied text. This influences subsequent generations, ensuring unique results for each interaction.

During text generation, the following characters can be chosen from:

• Emoji — Random emoji characters, from all faces and permutaions of skin color, to all flags and symbols, excluding provocative or NSFW emojis.

• ASCII — The usual printable characters, and some non-printable "invisible" tokens assigned for special purposes

• Unicode — Almost a million unicode characters from many languages, symbols, emojis, etc... Some unassigned

• Unicode PUA — Private Use Area unicode characters are characters reserved for private use, therefor have no inherent meaning

You have the freedom to select any combination of character sets, with each set having an equal chance of being picked during each character generation.

Chaos is a factor influencing the degree of variation in initial image grids. It can steer the generated results towards being unusual or unexpected.

Higher values of Chaos result in more varied and surprising initial image grids for each job run, while lower values yield more consistent outcomes.

Here's how Chaos impacts jobs:

• Low Chaos 0% — Slightly varied initial image grids per job run.

• High Chaos 50% — Highly varied and unpredictable initial image grids per job run.

• Very High Chaos 100% — Extremely varied initial image grids leading to unexpected results, challenging to achieve convergence.