Last updated - 18/06/2026
This is an offshoot of the main Illustrious model, which I've released separately due to the difference in architecture. V1.0 is based on V7.0 of the main series, using more-or-less the same dataset, while future versions may differ more.
This is also more of a test than a full release, as I've run into a lot of issues trying to get this model's performance up to standard. You can find a list of issues and other things I've noticed in the version description, which I'll update as I'm made aware of new things, so by all means let me know if you find something that isn't listed.
Additionally, I'd recommend using the tag "simple shading" as a negative in a lot of cases, which I've been using in my datasets for a while, but only now seems to have an effect. It refers to block colours, so using it as a negative should give more of a gradient-style lighting effect, which by extension tends to look more detailed.
I now have a Ko-fi for those that'd like to support my projects. While nothing there will remain exclusive, you'll get more substantial progress updates each month, and access to my own non-AI images at higher tiers.
Model description:
This is an attempt at an all-in-one lora for generating the monster girls of MGE, which allows for extensive customisation and feature-mixing. Refer to these articles for more information. Not everything works perfectly, so good results may require a bit of experimentation at times.
For updates on future versions, I generally post what I'm working on in my profile's announcement tab. It'll probably be MGE-related, but in addition to this one I occasionally make standalone concept-loras for monsters that would benefit from having one.
Background info:
"MGE" Refers to the monster girl encyclopedia book series by Kenkou Cross. For more information, see the MGE wiki here.
"Non-entertainment" usage:
Any "commercial projects" must be transformative, so no posting your results to Patreon and asking for money. Using it as part of a larger project is fine though, and I'm happy to help if you reach out with questions, or simply to tell me an interesting use-case you have. Bear in mind that if you're creating stuff linked to MGE specifically, they must also adhere to the creator of MGE's terms here.
Actual artists:
If any artists dislike the idea of their stuff being used for this (and you've drawn things from MGE), let me know and I'll avoid your work. I haven't done this preemptively due to the huge number of dead source links in the archive, but if you reach out (with some proof of identity to avoid trolls) I'll remove any of your content from my dataset/s.
Reuploads:
If you reupload this, please link both this model page and my taglist for ease of use. I'd also appreciate you letting me know where you've uploaded it, so I can keep track of where it's being used.
Description
This is more of a test version, which I've released to get some feedback in preparation for V7.1, so let me know how it performs. I'm still not sure how much I'll be able to fix, primarily because I'm not experienced enough with Anima to know what's causing the issues yet, but knowing what's broken will be useful nonetheless.
As far as my testing has shown, I'd say its performance is comparable to Illustrious V5.1, in terms of how it compares to its other V7.0 (Illustrious) counterpart, wherein there's some improvements but more broken things. I haven't had the time to do a full round of testing, and doing so would take longer than usual because Anima runs slower, so any feedback is greatly appreciated. I've started a list of thing's I've noticed below, which I'll update as-and-when I'm made aware of things:
A lot of artstyles don't work, though some of the ones that do perform surprisingly well (namely "Chibi", "Extra", "MGE" and "Pixel")
Horns are very inconsistent
Wings usually work
"Tentacle-legs" appears to break itself, "scylla" should still work
Normal lamias appear to be broken, but the bunyip works, strangely
Things involving "mixed textures" perform very well, like the roper's pink tentacles/green slime and the humpty-egg's shell and two slime colours
Slime-girls work quite well, though transparency is still inconsistent
"[Thing]-submerged" (with the al/liliraune) has regressed into generating external pools of liquid and/or water again
The barometz seems to work quite well proportion-wise, but the leaves don't appear to generate with backgrounds (seems to be negated by other green things, like grass)
"Gradient hair" and "[x] hair to [y] hair" both don't work, usually just making a patch of either colour instead
Clothing shows up more consistently, but is also prone to being more generic (e.g: a shoggoth will probably generate with a regular maid outfit instead of the "abyssal" one)
It appears to be even more sensitive than Illustrious in regards to body part mentions, so things like "boots" will usually force full-body shots
There's a funny bug with "baggy clothing", wherein it exaggerates the proportions of said clothing significantly. You'll see what I mean if you try and generate things like the Anubis or holstaur
Some monsters are completely broken (the dragon-zombie is the worst-performer I've found so far)
(more to follow)
FAQ
Comments (5)
Glad to see an Anima version, should be helpful in making offshoots of mge girls
This is awesome, was waiting for MGE collection for anima. Thank you bro
Nice, will be fun giving this one a whirl, and while I don't have much specific advice on how to improve the model's performance, I do know some tags that don't quite behave the way you'd expect them to in regards to how they're actually tagged on the boorus.
Examples:
Mermaids and amphibians: "webbed hands" either yields no results, or the model puts spider web on the hands instead. Odd solution for me was to change the tag to "membrane/skin between fingers" when prompting, meaning that's the key phrase to tag on in that regard.
Slime girls: "transparent skin/body" yields more balanced results than "see-through skin/body", both loraless and in fully trained model. Different weights produce different levels of transparency effect, but the head is more often than not excluded in either case. I suggest a experimenting with dedicated tags for the head/face/hair transparency.
Now, I don't know your precise workflow, I'm guessing you're segmenting up the training by datasets for each training run, correct? That's how I'd do it, at least. Either by batching them up by type (slimes, lamias, scyllas etc.) for a type focus for each training session, or specifically train each girl one by one, which may be time consuming, but probably worth it to ensure their designs will be learned properly.
Thanks, I appreciate the advice, I'll give them all a go next time.
In regards to training, I'm just using a single (big) dataset that gets updated as-and-when I get ideas. I have tags overlap as much as possible to try and make terms cross-reference each-other (e.g: using the same structure for "forearm-fur/scales"), meaning most changes are generalised and/or usually apply to a bunch of different monster categories. The dedicated loras are more species-focused though, obviously.
That, and I use Civitai's trainer, so I have to be careful with how much I train. It costs ~4000 buzz per run, so it'd get expensive quickly if I were to focus on/train each monster-category separately, potential benefits aside.
Here we go!