Question about Wildcards in prompts

[india7330]

Hello,
Following
https://github.com/vladmandic/sdnext/wiki/Wildcards
a few questions:

1. when talking about the weights in
   with {blonde:8:|brunette:4|red-head:2|purple highlights:1} hair,
   we are talking relative weights of what prompt replacement will be choosen, right?
   This has nothing to do with prompt weighting for the model, right?

2. From my understanding, token count should NOT change if an expression is replaced with a braced wildcard?
   Because the wildcard should replace text before handed over to token engine.
   But it does significantly. Why?

3. When using wildcards in file, like in example:
   red
   green
   blue
   , is there possibility to add weight as well?

[vladmandic]

1 - correct
2 - if you're talking about token counter in the ui, its because it cannot run full prompt parsing in real time, it runs on exact prompt as-is
if wildcards do replacements, yes, it will change the prompt and actual token count will be different
3 - no as it would mean that entire wildcards file would need to be fully evaluated before picking an entry and some of them can be massive

[india7330]

Thanks!
Follow-up questions:

4. In {red:0.6|blue}, the default when not specified (for blue) is 1, right?

5. A single value, {hat:0.5} doesn't make any sense, right? It still has 100% weight, irrespective of value? Or will it be selected 50% of time in this example?

6. If i want hat 50% of time, an alternative writing would be {hat|}, i.e. the "empty slot" is considered one of possible options?

(maybe you could update documentation with these answers)

[vladmandic]

default is 1 and yes, single value doesnt make sense. and yes, you can use empty slot as possible option.