UPDATE! (20.01.2025)
v1.0 Illustrious-XL v0.1
UPDATE! (10.11.2024)
v1.0 NoobAI-XL EPS 1.0
UPDATE! (15.07.2024)
v1.0 Pony XL v6 - A small update to the Pony version, Almost the same dataset as in the previous version, but with better upscaling and captions have been edited, the sw_jessicadrew token was changed to jessicadrew
UPDATE! (12.03.2024)
SDXL, Animagine XL V3 Base, Animagine XL V3
UPDATE! (03.02.2024)
(Pony XL) Only for ponyDiffusionV6XL_v6
UPDATE! (25.12.2023)
v5.0 - 40 New images & realictic images
UPDATE! (15.12.2023)
v4.0 - New Dataset & Improved Training
v4.5 - Added Tokens: white lens, covered eyes, no mask (this version is no better than v4.0, but it works differently)
v4.5 Megre - animefull-final-pruned-fp16 & v1-5-pruned-emaonly (This version works a little better with realistic models)
UPDATE! (25.06.2023)
the models is a bit overcooked so I suggest you use Dynamic Thresholding (CFG Scale Fix)
The v1.0 & v2.0 images were generated with VAE (vae-ft-mse-840000-ema-pruned) so they look a bit overcooked
I still recommend using Dynamic Thresholding (CFG Scale Fix) for the (sd-v1.5) version
CGF Scale: 4 & Mimic CFG Scale: 8
https://github.com/mcmonkeyprojects/sd-dynamic-thresholding
v3.0 works slightly better with lenses on the mask and is not overcooked on the (sd-v1.5) version
LoRA Weight:
v1.0 & v2.0 [0.5]
v3.0 [0.5-0.7]
v1.0 & v 2.0
CGF Scale: 5-6
v3.0
CGF Scale: 7-8
Resolution: 512x768
Description
Repeats: 10
Epoch: 5
Image Count: 100
Network Rank: 32
Network Alpha: 16
Batch size: 2
Clip skip: 1
Training steps: 2500
Resolution: ??0x960
Train Resolution: 1024x1024
Model: animagineXLV3_v30
Captioning: WD14 Captioning + Edit (SmilingWolf/wd-v1-4-convnextv2-tagger-v2)
Trigger Words: sw_jessicadrew
LoRA Weight: [0.8-1]
FAQ
Details
Files
Available On (20 platforms)
Same model published on other platforms. May have additional downloads or version variants.