Sign In

[LECO] Monster meat

143
689
3.9k
30
Updated: Mar 1, 2025
objectsdarkfoodmeat
Verified:
SafeTensor
Type
LoRA
Stats
124
4
1
Reviews
Published
Feb 28, 2025
Base Model
Illustrious
Usage Tips
Strength: 1
Trigger Words
meat
Training Images
Download
Hash
AutoV2
F8BAD5A439
default creator card background decoration
toretate

This LoRA is a demonic meat.

use for dark fantasy meal sean.

for Illustrious v0.1

This is a demonic foods (re create for Illustrious v0.1)

This is LECO, good feeling weight is 2.0 ~2.5.

trigger word: meat

wieght: 1 ~ 2

LECO Parameters

- target: "meat"
  positive: "meat, monster, purple, tentacles, multi colored, eyes, blue, yellow, black, insect, dripping, glowing"
  action: "enhance"
  guidance_scale: 2.0
  resolution: 768
  dynamic_resolution: true
  batch_size: 3

below: SuperMerger LoRA merge parameters

ratio: 2
Blocs: 1,0,0,0,0,0,0,0,0,0,1,1,1,1,1,1,1,1,1,1

Be careful when setting Strength to 2 or higher, as it tends to produce significantly darker images.

SD1 版のものを Illustrious v0.1 で作り直したもの

LECOのパラメータ

- target: "meat"
  positive: "meat, monster, purple, tentacles, multi colored, eyes, blue, yellow, black, insect, dripping, glowing"
  action: "enhance"
  guidance_scale: 2.0
  resolution: 768
  dynamic_resolution: true
  batch_size: 3

また、SuperMergerで強度などを補正済みです

ratio: 2
Blocs: 1,0,0,0,0,0,0,0,0,0,1,1,1,1,1,1,1,1,1,1

Strength を 2以上にするとかなりダーク寄りの絵が出てしまうので注意?

ダークファンタジーの食事風景にでもご利用ください


for SD 1

This is a demonic foods.

This is LECO, good feeling weight is 2.0 ~2.5.

If you are going to use it with human, use Regional Prompter or Regional tools. ( If not used with the regional tool, it would often fill the entire image)

trigger word: meat

wieght: 2.0 ~ 2.5 (if Regional tool, use 1~2)

LECO parameters

- target: "meat"
  positive: "meat, monster, tentacles, purple, blue, multi colored, insect, sticky"
  action: "enhance"
  guidance_scale: 1.0
  resolution: 768
  dynamic_resolution: false
  batch_size: 2
- train:
  precision: "bfloat16"
  noise_scheduler: "ddim"
  iterations: 200
  lr: 1
  optimizer: "Prodigy"
  lr_scheduler: "constant"
  max_denoising_steps: 20
  network:
  type: "lierla"
  rank: 4
  alpha: 1.0
  training_method: "full"