Since all of your generations used a weight of :1, I believe this might be the reason. You should try and lower it to around :0.5~0.8 for better results.

@Meji i never lower weights, because i don't want to get checkpoints mix in their style. and there is zero reason to lower weight anyway. if you take a look on my profile, you will see that i only work this way to see what loras really are made of.

@stapfschuh Then uh... no offense, but how do you handle lora's that utilize a spectrum of negative -8 to +8? https://civitai.com/models/128553/eye-size-slider-lora 😉
In all seriousness~ 🤔 I understand the reasoning of testing "pure Lora mettle" for your own verification purposes, but all generations are inevitably influenced by their checkpoint/seed. That's why most Lora will be deliberately strengthened... because some elements (nudity)will be more prominent in a checkpoint than others~(elf). You can even finesse a lora by an additional decimal. 0.72 for example, which means it can be attenuated to different checkpoint environments... so why put up the "half-assed" generations? So long as the meta-data/prompt is available, it's not dishonest... but you seem like you're actively making the lora look bad.

I do really appreciate the "loi" tip though! I'll be including it as a note for myself once I'm using this Lora.

@taiconan i simply like to tickle out those images on what these loras ar trained on and see if i get the correct character that it has to be. and it always works.

and about loras with different weights.. these are mostly sliders and i know how to use them in a proper way if i would need them.

that now was not meant in an offensive way, i just replied.