Not recommended for realistic model
Main tag : mtu virus,
Secondary tag : multiple views,
LoRA str : 1.0-1.2
Denoising str : 0.25
Training Powershell info :
DIM/Alpha : 8/8
Train batch size : 2
resolution : 768
mix/save_precision = BF16
optimizer_type = Lion
GPU : RTX 4060 16GB 160W
Torch 2.51 + cu12.1 + Xformers + gradient_checkpointing
Vram used : 11.3GB
Speed : 1.47sec/it
Description
FAQ
Comments (16)
Is there any problem with yesterday's one
TOA Image
@Gtonero My English is not very good. Can you explain it in detail, preferably without abbreviations
@teenager Notes that have been reported are
Images involve too many children.
@Gtonero Also, let me ask how to adjust this ratio. I have tried everything from 0.1, 0.6, and 1.0, and none of them look like a demonstration image. I used the same tag you demonstrated
@teenager ratio 1:1
@Gtonero Weight？
@Gtonero What else is the model used for??? The real person model I used
@teenager weight 1
model anime not sure if this will work for realistic models
@Gtonero Okay, so far, most real models have failed. I'll try again later. If possible, could you also create a Lora like this for a real model
I think it's a pretty good LORA, but I think it's an improvement point because if you wear pantyhose, it makes it less meaningful to peek.
pantyhose while train. I think image are not enough
How do you peek from behind? I always peek from the front
You put (ass:1.2) in prompt? Duh.
Why is character painting also viewed from below? Why does character painting look from the front instead of the bottom
install locon Can cause other lora lose efficacy