Skip to content

Latest commit

 

History

History
46 lines (34 loc) · 2.55 KB

Reference.md

File metadata and controls

46 lines (34 loc) · 2.55 KB
Raw

This document is a stub. You can contribute and help this project by adding more details and examples to this document. Thanks!

Prompt Syntax

Attention Weights

Append a word or phrase with - or +, or a weight between 0 and 2 (1 is default), to decrease or increase the importance of that word/phrase in the generated image.

You can assign weights to multiple words by using parentheses. For e.g. a man (picking apricots)1.5 or a man (picking apricots)++

You can add more + or - symbols to increase/decrease the weight further. For e.g. apricot++ has more importance than apricot+, and apricot-- has less importance than apricot-. There's no limit to how many + or - symbols you can use.

You can also use a number to assign an exact weight to a word/phrase:

  • A weight between 0 and 1.0 reduces the importance of the token. For e.g. (apricots)0.5 reduces the importance of apricots to half.
  • A weight between 1.0 and 2 increases the importance of the token. For e.g. (apricots)1.5 increases the importance of apricots by 1.5 times.

+ is essentially a weight of 1.1, and - is essentially a weight of 0.9.

More examples:

  • nesting: a tall thin man (picking apricots+)++ (apricots effectively gets +++)
  • single words without parentheses: a tall thin man picking apricots+
  • single or multiple words with parentheses: a tall thin man (picking apricots)+ a tall thin man picking (apricots)- a tall thin man (picking apricots)-
  • more effect with more symbols: a tall thin man (picking apricots)++, and a tall thin man (picking apricots)+++
  • all of the above with explicit numbers: a tall thin man picking (apricots)1.1 a tall thin man (picking (apricots)1.3)1.1. (+ is equivalent to 1.1, ++ is 1.1 x 1.1, +++ is 1.1 x 1.1 x 1.1, etc; - means 0.9, -- means 0.9 x 0.9, etc.)

Blending between prompts

You can blend between concepts in the prompt by using the .blend() function.

For e.g.: ("blue sphere", "red cube").blend(0.25,0.75)

This will tell the sampler to blend 25% of the concept of a blue sphere with 75% of the concept of a red cube. The blend weights can use any combination of integers and floating point numbers.

Escaping parentheses () and speech marks ""

If the prompt you are using has parentheses () or speech marks "" as part of its syntax, you will need to "escape" these using a backslash, so that(my_keyword) becomes \(my_keyword\). Otherwise, the prompt parser will attempt to interpret the parentheses as part of the prompt syntax and it will get confused.