This is a subvariant based on my General Use model Midgard made for Animals/ Creatures and Furry Art but can do way more.
This model can also used for:
Photorealistic
Digital Art
Drawings
Anime
The Model can do SFW and NSFW
This Model contains also parts of the Model Pony Diffusion V6 XL , my Ratatoskr and the Furry Enhancer .
This model is way more Realistic than the original . also the prompting is more easy it can prompted in 3 ways : Natural Language, E621, and the Pony prompting, also mixed is possible .
V2.3
First Release Version recommended for General use include Photorealistic
V3.0
more creative
made for semi realistic and artwork and anime but still can do photorealistic
better in nsfw
optimized for furry artwork
optimized for the furry enhancer V6
This Version Contains parts of the model Compassmix XL made by frosting and Mjölnir
V3.2
combines photorealistic with creativity
more fine detailed
baked in the furry enhancer v6.1+ unreleased version
better with artworks
better nsfw
Note: the model is a bit more horny if this is a problem i recommend still using V3.0
This version contains also some part of the following model: Demon Core V4.1 Midgard Beast
Note: the model has a hybrid between SDXL Lightning and pony so it's unclear what type this model actually is.
for Settings i recommend looking on the original Midgard Model because this model very similar in settings.
Made by Freek22
Please to support this model give a like also check out my other models too ;-)
Description
combines photorealistic with creativity
more fine detailed
baked in the furry enhancer v6.1+ unreleased version
better with artworks
better nsfw
Note: the model is a bit more horny if this is a problem i recommend still using V3.0
FAQ
Comments (16)
Incredible good model for photorealistic anthros. I created a large batch of images of anthropomorphic vixens at the beach and it was like jaw dropping realistic and high quality. Highly recommend this checkpoint (and Ratatoskr)
that's really good to hear have fun with the model
At the moment there are some very strong pony/pdxl models out there.And I am curious at what content this model is shining,besides that furry-stuff.
So what kind of Tits and Tops,Hits and Hots,Flips and Flops,Sh..ts and Shots? :-)
what this model can do is prompted in 3 ways (sdxl, e621, and pony) also the model can be run similar like an lightning and hyper model but the usual sdxl step count is also fine. For the contend the best is to look through the example pictures. the Model has a high range from drawing to realistic. the model is based on the general use model midgard so it has a huge variety , from scifi , fantasy , still, animal. etc. It's for sure not most realistic model i made but has its own style ;-)
@freek22 Thank you very much for your information. Yeah these pony prompts are one reason I am a little shy,at the moment,getting into this kind of models.And the usual furry and anime parts are not really my cup of tea at the moment. On the other hand - high realistic humanoid/human like persons/creatures for character/story-telling/evolution are very welcome.And it seems to me that pony/pdxl is very suitable for this.
@mscorpion7285 it is suitable for this kind of images, it has a strong bias towards anthro and fantasy creatures but if you need a realist or an art model i have made other models that are better in that area than this " pony " model like the normal midgard, mjölnir, sleipnir for examples
I played around with tons of settings for v3.2 in my ComfyUI workflow. I found that DPM++ SDE Karras works best for photo-realistic, 3 or 4 CFG and 14 to 16 steps.
For a smoother anime or illustration look, try out DPM++ 2M SDE Karras, DDIM, or LMS (very smooth).
Additionally, Pony score_9 etc. tags will only add noise, use normal quality tags for SDXL Lightning.
yes dpm++ sde is good with it the score taggs are not necessary for the model
Guys i can tell you one thing about this Checkpoint.
it is Amazing maybe the Best Checkpoint what you can have in SDXL
The Option with Ponys Score prompting for quality and you can mix it with natural Language with the typical tag prompting.
I use Pony Base Models for my Images but if i wanted to inpainted my creations i use Midgard because Fooocus A.i can only use SDXL Base Models for the inpaint mode and that is the best Combination what you can have.
It is For me as non English user much easier describe my scenery
thank you for it @freek22
Hello! I made a comparison of Pony models. 1 test prompt so far. More ongoing.
https://civitai.com/articles/6491
Comments / suggestions are welcome.
what i can say this model use a lower cfg scale than other pony models so 3-4 and dpm++sde are working fine also the model don't need the score taggs to work.
btw it's really nice that you do such time consuming work to compare models i know that is not easy .
@freek22 I hope someone will use this test to make even better model by taking the best parts from tested models.
"The spice must flow..."
@dioxidin with only mixing model together you will not get a good model for sure.
You need finetuned elements etc. Otherwise you end up with an average model overall without any custom outputs and normally boring.
This why I like that huge variety in the models. I like your descriptions about the models it can help others to find the "right" model for the expected porpoises. This model is for example finetuned for anthro and animals, others prefer realism etc.
Compared to NoobAI, how many tags does Midgard cover? Others have made 500K wildcard tables, so would it cover like a large chunk of them?
possible but the model is way older than NoobAI its better to compare with my new model too (ratatoskr helheim)
Details
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.