Sign In

Zelda CDI Style LoRA

355
2.1k
10
Verified:
SafeTensor
Type
LoRA
Stats
2,057
Reviews
Published
Jan 31, 2023
Base Model
SD 1.4
Training
Steps: 4,840
Epochs: 5
Training Images
Download
Hash
AutoV2
172EA216B3
default creator card background decoration
tsu's Avatar
tsu

Zelda CDI style LoRA

This is a LoRA, this mean that you need to update Voldy/AUTOMATIC1111's webui to use it (use git pull) .

That LoRA does not have a particular trigger (except the default <lora:LoRaName:1>).

You can use it at weight 1.0 but since it's a 256 dim model, there is a possibility that it overfit in some situation and if it happens to overfit you can lower the weight to 0.9 or 0.8 .

121 images, 8 repeats,5 epoch,4840 steps
Learning rate: 1e-4
Text encoder learning rate: 5e-5 
Unet learning rate: 1e-4 
max bucket resolution: 512 
clip skip: 2 
dim/alpha: 256 
batch size=1 
Base model: a Notorious AnythingV3 Important ancestor (you get it?) this mean that this LoRA work on any model derived from that particular model like AnythingV3/4/4.5, AOM 1/2,etc...

All characters are owned by Nintendo even if they pretend that this game never existed.