V2.0_C2
The photo editing tool SDforge
step: 30
cfg:5
Sampler: DPM++ 2M SDE
Schedule type: Karras
checkpoint:
waiMatureIllustrious_v10
Strength recommendation:
<lora:Kiriko_Hattori_SDXL_IL_V2N_C2.1:0.7>~<lora:Kiriko_Hattori_SDXL_IL_V2N_C2.1:1>
Prompt：
ha2to1,1girl,solo,brown hair,long hair,ponytail,hairpin,headband,green eyes,lipstick,purple bodysuit,exposed breasts,gold trim,japanese armor,lace-up,breastplate,midriff,arm armor,shoulder blades,leg armor,high heels,p sandals,
V2.0_C1
The photo editing tool SDforge
step: 30
cfg:5
Sampler: DPM++ 2M SDE
Schedule type: Karras
checkpoint:
waiMatureIllustrious_v10
Strength recommendation:
<lora:Kiriko Hattori_SDXL_IL_V2(N)_C1:0.6>~<lora:Kiriko Hattori_SDXL_IL_V2(N)_C1:1>
Prompt：
ha2to1,1girl,solo,brown hair,long hair,ponytail,green eyes,round eyewear,lipstick,red choker,bare shoulders,clothing cutout,cleavage,large breasts,red japanese clothes,sakura,lace-up,waistband,wrist guards,fishnet thighhighs,high heels,r sandals,
The photo editing tool SDforge
step: 30
cfg: 3
Sampler: DPM++ 2M SDE
Schedule type: Karras
checkpoint:
waiMatureIllustrious_v10
Prompt：
1.<lora:Kiriko_SDXL_IL_V1:0.9>,ha2to1,1girl,solo,brown hair,long hair,hair between eyes,ponytail,hairpin,black headband,green eyes,lipstick,large breasts,purple bodysuit,(exposed breasts:0.8),gold trim,(japanese armor),lace-up,breastplate,arm armor,shoulder blades,purple gloves,leg armor,high heels,p sandals,
2.<lora:Kiriko_SDXL_IL_V1:0.6>,ha2to1,1girl,solo,brown hair,long hair,ponytail,green eyes,round eyewear,lipstick,red choker,cleavage,large breasts,((sakura)),red ninja,wide sleeves,wrist sleeves,bare shoulders,lace-up,wrist guards,short kosode,((fishnet thighhighs)),high heels,r sandal,
PS:
The special cutting of this character's costume is beyond my ability to describe. The effects presented by the two sets of clothes are unstable, and each set has different effects with varying weights. The controllability is poor. Only when used together with the pose LORA model can it be barely acceptable. Please explore on your own. I will look into the methods for optimization later.
Description
Only the model of one set of clothing will be re-trained, and then the LORA for the other set will also be created.The model was re-trained using a new training method. The restoration degree of the new model is very high. However, there is a problem. If handwritten prompts are used, the effect is not very good. The randomness and flexibility of the image are poor. Some actions will not be responded to by the model if the prompt words are not clearly written or the weight of the prompt words is not increased. It is best to use the corresponding action-based LORA model in combination.
FAQ
Details
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.