In Stable Diffusion generation, there is a feature called wildcard, which randomly picks registered words to create prompts.
In the prompt below, the words surrounded by double underscores (__) are wildcards.
__LoRA__, upper body, (nak**:1.3), __smile__, __pose__, simple background, white background
I tried generating with LoRA, smile (expression), and pose as wildcards, while keeping everything else uniform.
LoRA is an auxiliary model that has learned the characteristics of a character (in this case).
There are many good LoRAs for Tifa and Aerith, so there are plenty of images.
Elena and Kyrie are a bit iffy, but there are fewer issues.
Yuffie is problematic with too many issues, so there are fewer images.
NMR
2024-07-28 18:26:17 +0000 UTCSkeletron27
2024-07-27 18:11:26 +0000 UTCEather989
2024-07-27 11:46:28 +0000 UTC森のくまさん
2024-07-27 11:18:29 +0000 UTCProtoedj
2024-07-27 11:11:27 +0000 UTC