Yo. Its been a while i was having a very hard time getting results better than AIOMonstergirlsV3. I had trained merged trained merged again and so I took a break but now im back at it. This version has improved the quality of the image by a lot and also added two more monster girls that where requested.
when training theses version i balanced the dataset making it so Lamia wasn't over-trained and Scylla under-trained. Also I refined the dataset by choosing quality over quantity and trimmed down on the captions. The style has change a bit again going for a more 2D look but with prompting and CFG you can get that 2.5D look again
AIOMonstergirlsV4: The main checkpoint which is just a improvement from the base model with good overall concept comprehension
AIOV4Lora/lyCORIS: An extracted LORA from V4 about 90% accurate to the full checkpoint you will see some discrepancies it it about 340MB as its trying to put a full checkpoint and all the monster girls into a LORA
one thing that has changed is I still recommend ADetailer but Dynamic Thresholding isnt really needed anymore and great results can be had with low steps and little tweaking
AIOMonstergirlsV4 new monster girls:
Frog girl
Mantis girl
The LoRA/LoCON version is a LoRA not a checkpoint
Feel free to comment what other monster girls you want in for the next version. Most monster girls from Monster Girl Encyclopedia are somewhat trained but would need more training to get the quality i want. Have fun generating
Description
Complete change of style as I really like the saturation and shading of this version more
FAQ
Comments (27)
can you make your model merge with the expline v2 model i found it to color very well fantasy i used your lora to use on that model it was very bad
I can't find a model named Expline is it on Civitai?
@pjaama https://civitai.com/models/44150 here it is I hope it can match your model
Any chance you'd make the satyr it's own LORA?
I can try but im not great at making LORAs :)
alraune pls
also humanoid versions i think nymph/dryad for example
Would you be able to extract a LoRA out of V3? It seems like the LoRA isn't quite V3.
Excellent model, looking forward to your future updates❤
May I request Empusa/Mantis Girls, if it hasn't already been add?
lamia always has redundant tails and discontinuous tails
did some test with various prompts/settings/seeds/etc. (checkpoint v3)
good results
arachne, black harpy, centaur, gazer, harpy, jabberwock, lamia, mermaid, merrow, mershark, roper, succubus, undine, unicorn, zombie
model know this but it might be differ from what do you expect
alraune, ant, arachne, anubis, apophis, atlach-nacha, bicorn, cheshire cat, crow tengu, dark matter, giant ant, golem, hellhound, high orc, holstaur, imp, jiangshi, khepri, krampus, large mouse, lava golem, lich, mad hatter, mandragora, mantis, march hare, mindflayer, mucus toad, nereid, orc, pharao, red oni, satyros, scylla, sea slime, shoggoth, skeleton, slime, slime carrier, soldier beetle, sylph, ushi-oni, vamp, mosquito, vampire, werebat, wererabbit, weresheep, werewolf, wight, wurm, wyvern
and.. this is just my luck or whatever, beware of following words or put l-word and s-word to negative
dorome, dwarf, elf fairy, genie lilim, otohime, queen slime, ryu
Im begging you please, Yeti, mothman, and holastar
Medusa or other gorgon is a good option
I've never seen such a huge (2GB) Lora before. Even with some "megapack" Loras with multiple characters and tons of clothes. Why do you think your Lora got so big ?
This is not a LoRA, this is a checkpoint.
This model keeps on surprising me with all the hidden styles trained inside it. It also works really great for its intended purpose (unless you want transparent slime girls, in which case you still need a separate LoRA but at least it works).
How do you get slime girls with different colors?
Just specify the color? Seemed to work as expected. I made these with this model, then used a different checkpoint to convert to that retro style: https://civitai.com/posts/465768
First of all, I would like to express my gratitude for providing this model. I am thoroughly enjoying using it.
With that said, I have a request. I would love to see a girl modeled after a spider with three arms or a girl inspired by a praying mantis.
Additionally, it would be helpful to have LoRA files for versions 3 and beyond.
I am rooting for you and wish you all the best in your future endeavors!
(Translation from Japanese' in English)
is it possible to put the top half of a girl on the head of a monster with spikes?
hmm maybe monster girl versions of certain lovecraftian creatures given some monster girl versions of them do look good despite the original
also maybe dryad/nymph basically one or 2 humanoid versions of the alraune
angel girls
different demon girls like succubus or oni
dragon/lizard girls (more human/elf like not anthro)
Hey I can not quite get your guyss results, I am useing the dynamic thresholding right but my images still look off are you guys using loras or textinversions or something? Can not replicate most of the pics :c
Whenever I try to run this as a Lora, I get this error:
RuntimeError: Expected all tensors to be on the same device, but found at least two devices, cuda:0 and cpu! (when checking argument for argument mat2 in method wrapper_CUDA_mm)
I've tried removing and re-downloading it several times, and going through both the a1111 additional networks and the native Lora system and they both give this error.
I've never seen this before for any other Lora.
Edit: For some reason, the very next time I tried the Lora after writing comment, it didn't throw the error, but it also didn't seem to do anything to the picture (I used it via the native lora hash, which put <lora:aiomonstergirls_aiov2LoraLocon:1> in the prompt section)
Edit Edit: When I tried doing it through the additional networks extension, it didn't work, and when I went back to using the hash, it also no longer worked. Bizarre.
This is a full model my guy, not a Lora.
@kawaiibully One of them is a Lora/Locon.
A great model for monster girls. The only downside is that it requires very fine tuning for quality results. Difficult after other models
Details
Files
Available On (2 platforms)
Same model published on other platforms. May have additional downloads or version variants.







