
This lora is meant to generate a particular type of heels as shown in pictures. I have a shoe fetish for this particular type of heel and made a lora for it in SD 1.5. SD1.5 is so outdated now so I wanted to finally try remaking it for SDXL. Every time I tried to prompt for it in SDXL models, no sdxl checkpoint in recent times has been able to generate this particular type of shoe consistently and it was frustrating so I finally made a lora for it to make it very consistent.
Use a lower weight like 0.3-0.6. Most modern checkpoints already have a strong understanding of the shoe, it just needs to be nudged/pushed in the right direction with this lora to make it consistent; it does not need to be forced aka a weight of 1.0. The training data is of only on a foot wearing the shoe, so if you put the weight high you may get pictures focused solely on the shoe.
There are various colored shoes in the training data, but mostly black and clear.
The full training data caption is "female feet wearing platform heels, mules, high heels, muules."
Trigger word - muules
NOTE: I cannot fix the wonky toes, thats just a major limitation of SDXL as a whole, along with fingers.
描述:
first version
训练词语:
名称: 866512_training_data.zip
大小 (KB): 220666
类型: Training Data
Pickle 扫描结果: Success
Pickle 扫描信息: No Pickle imports
病毒扫描结果: Success
名称: Platform_mules-000009.safetensors
大小 (KB): 223102
类型: Model
Pickle 扫描结果: Success
Pickle 扫描信息: No Pickle imports
病毒扫描结果: Success