The LHC (Large Heap o' Chuubas) is aiming to be the model for all of your VTuber needs. There are other minor goals, like improving aesthetics, backgrounds and ???????, but the main goal is to offer a lora-less option for generating VTubers.
Alpha V0.3.1
Due to some mistakes during the training of alpha v0.3, the model has diverged significantly from NoobAI. Nonetheless, it is a capable model with good understanding of most of the 79 trained vtubers and a passable one for the rest. For an overview refer to:
https://huggingface.co/Jyrrata/LHC_XL/blob/main/characters/alpha03.txt
and https://civitai.com/posts/9579061 for a visual guide on basic character comprehension of the two v0.3 models. Many characters work with only their activation tag, though some require a little or a lot of additional tags to work.
Alpha V0.3 and V0.3.1 were trained on NoobAI-XL V-Pred-0.6 version.
If you want to use V0.3, it can be found here: https://huggingface.co/Jyrrata/LHC_XL/blob/main/alpha/v03/LHC_alphav03-vpred.safetensors
Additionally, there is also an eps version and a version trained on rouwei-vpred of intermediate datasets in that huggingface repo. Refer to the character .txt files for overview of the v0.2.5 knowledge.
Alpha V0.2
Same general approach as v0.1, however the dataset has been expanded by 10 additional vtubers for a total of 28 now, and the final two epochs include an experimental dataset of 1200 images covering a wide base of concepts intended to realign and improve the model aesthetically.
Included vtubers this time are:
aradia ravencroft
bon \(vtuber\)
coni confetti
dizzy dokuro
dooby \(vtuber\)
haruka karibu
juniper actias
kogenei niko
malpha ravencroft
mamarissa
michi mochievee
rindo chihaya
rin penrose
atlas anarchy
dr.nova\(e\)
eimi isami
isaki riona
jaiden animations
juna unagi
kikirara vivi
mizumiya su
trickywi
tsukinoki tirol
alias nono
biscotti \(vtuber\)
mono monet
rem kanashibari
yumi the witch
In addition to adding new ones, the datasets for some of the old ones have been redone, especially trickywi, juna unagi and juniper actias. Juniper has also gotten two new tags, juniper actias \(new design\) and juniper actias \(old design\), which tries to seperate her models into two distinct phases. This is experimental and might not be carried forward to future versions.
A showcase of the base character tag understanding is here. Some vtubers don't work with only their character tag, instead you will need additional descriptive tags.
Alpha V0.1
This model is currently still in alpha. The current state is not indicative of all future capabilities, but rather just a proof of concept.
A basic test model, with nice results nonetheless. Trained on roughly 1000 images featuring mostly 18 vtubers that the base NoobAI model did not know well. This model is based on the NoobAIXL v-pred-0.5-version model.
As a V-pred model, this model will not work in all WebUIs, but only those that have implemented vpred sampling. The necessary state dicts of the model have been set for UIs like Comfy and ReForge to set the required settings automatically. If not, it is necessary to activate v-pred sampling and it is recommended to turn on ztsnr as well.
The newly added/enhanced vtubers are (listed by their trained tags):
Aradia Ravencroft
Malpha Ravencroft
Mamarissa
Koganei Niko
Rindo Chihaya
Mizumiya Su
Isaki Riona
Kikirara Vivi
Coni Confetti
Dizzy Dokuro
Dooby (Vtuber)
Haruka Karibu
Juna Unagi
Juniper Actias
Michi Mochievee
Rin Penrose
Trickywi
Jaiden Animations
Additionally included were especially Nerissa Ravencroft and Vienna (Vtuber), as well as many images featuring 2 or more characters at once.
For a showcase of the base character comprehension, check out this post.
Recommended Settings:
Sampler: Euler
CFG: 4-5
Steps: 25+
Training Details:
Trained as a full dimension LoKr, based on the methodology of the KohakuXL series, with the Lycoris settings found here.
Specific parameters:
Dataset: 1035 images
Batchsize: 2
Gradient Accumulation: 4
Training steps: ~6400
Training Epochs: ~50
Unet LR: 3e-5 (lowered to 2e-5 for the last 12 epochs)
TE LR: 2e-5 (lowered to 1e-5 for the last 12 epochs)
Optimizer: AdamW 8-bit
Constant scheduler
Special Thanks:
kblueleaf (Kohaku Blueleaf): for the Lycoris library and the resources on finetuning via LoKr
OnomaAI & Laxhar Dream Lab: for amazing base models
kohya-ss: for sd-scripts