
Somewhat trained on Kkasi's work: https://twitter.com/kkasi00, though a bit different.
Kkasi is a giantess artist, so this will sometimes be biased towards "from below" angles, and certain poses, and also may generate tiny people and background buildings. Negative prompting it out works if that's not your thing.
I use this more often in combination with other Lora models, so I tend to use it 0.5, but 0.7 works fine by itself.
The examples I provide use my own mixed model: https://civitai.com/models/61700/anythingorange03mix
The "Before" images do not use the LoRA model, and are to differentiate the LoRA from my style of model and prompting. The similar image after is what happens when you put it through img2img with the same prompt, just with the LoRA added (and around 0.5-0.7 denoising). The difference is subtle sometimes, but I personally think it adds just the right amount of spiciness to the image.
描述:
训练词语: Kkasi
名称: kkasi_v1.0.safetensors
大小 (KB): 147571
类型: Model
Pickle 扫描结果: Success
Pickle 扫描信息: No Pickle imports
病毒扫描结果: Success