Pretty sure I've experimented enough to see a gap in the processing. A negative prompt like this, (easynegative:6), should understandably make an ugly image because of over emphasis on the embedding. However, with smZ CLIPTextEncode I see no difference between (easynegative) and (easynegative:6) or (easynegative:6.0) or (embedding:easynegative:6.0). I can only get the weight of an embedding to change if I do something like ((((easynegative)))). It seems like the behavior is the same weather used in positive or negative conditioning.
Pretty sure I've experimented enough to see a gap in the processing. A negative prompt like this,
(easynegative:6)
, should understandably make an ugly image because of over emphasis on the embedding. However, with smZ CLIPTextEncode I see no difference between(easynegative)
and(easynegative:6)
or(easynegative:6.0)
or(embedding:easynegative:6.0)
. I can only get the weight of an embedding to change if I do something like((((easynegative))))
. It seems like the behavior is the same weather used in positive or negative conditioning.