HIGHLY RECOMMEND ZoinkScoob INSTEAD, IT HAS VORE BUILT INTO THE TRAINING SET MAKING IT A LOT BETTER
Trained on a variety of vore concepts.
NOTE: Needs upscaling for the prey to be seen clearly.
*Very* picky with tag orders and lora strength.
Apologies about the large lora size, I couldn't make it smaller without losing quality.
!!IMPORTANT!! Use lora strength of 0.75 !!IMPORTANT!!
DOES NOT WORK ON SEAART.
GROSS WARNING: Sometimes if the cfg/tags/lora strength is set wrong disposal concepts may bleed over into other images, so you may see scat/bones even if tagged with soft vore. You may want to put them in negatives.
If you use the 1.0 lora strength your image will be immediately fried. You can test with different strengths, but going closer to 0.8 and up seems to start reducing quality, adherence to style and general cohesion of the image.
Always use (anthro pred) in the tags, it seems to increase quality.
You can find the custom tags in the yaml file. It's really just a txt, but I couldn't upload it as a txt so I had to do it that way, open it in notepad. I recommend copy pasting them and adding them to \webui-forge\webui\extensions\a1111-sd-webui-tagcomplete\tags\e621.csv
Generating at 1280x1280 seems to improve quality.
More voreconcepts tags is better, describing the image. If you only have a couple the quality will be worse.
Adding weights to the actions and putting them together like for example ((voreconcepts oral vore:1.3), (voreconcepts vore:1.2)) helps with getting the image more likely to show the concept, as well as increasing quality. The lora likes to not always follow the prompt, so if it's not working, you can add or remove descriptors (saliva, wet, abdominal bulge, etc), add or remove voreconcepts tags, increase weights (1 -> 1.1, 1.2), change the lora weight slightly (0.75 -> 0.7, 0.8).
Put voreconcepts tags together to increase coherency. anthro pred, (voreconcepts vore, voreconcepts oral vore, voreconcepts almost fully inside, voreconcepts partially inside)
Note: Some of the images were generated on less trained versions of voreconceptsmix, so copying the prompt may not work.
Has:
Oral Vore
Anal Vore
Cock Vore
Unbirthing (A little finnicky atm, I will increase the dataset for next version.)
Imminent Vore
Extreme Size Difference
Digestion
Fatal Vore
Internal/Cutaway
After Vore
Burping
Disposal
Description
Based on v1.0
FAQ
Comments (11)
🤩 Impressive! Watching with intrigue, but prohibitively large (for me) in it's current state. I'm willing to bet you could parse this into a couple separated loras (Like the belly/ball impressions as one, and the one with a head sticking out of an orifice being another, etc. etc.)
Thanks! Sorry about the large filesize, when I decreased it with the lora resizers the model seemed to drop a lot in quality. I intend to make some seperate loras later, but it'll be a bit of time since making this model took a while.
I went ahead and uploaded the rescaled version (200mb) anyway, I figure it's better to have it available even if I can't guarantee it'll work ^^
no unbirth?
It is very RNG atm, as I didn't put enough in the dataset. Will add more for next version. But it shows up if you're willing to mess around with the prompt enough. I added an example to the post's images.
On the image idk why it says I used the CV lora, I used my own lora so it's just a bug from civitai's side.
Work with pony models?
Not at the moment. Planning on making one later.
@Hallancer It would be very helpful if you could state in the information which model this is for, rather than "Other". I take it by the fact that it doesn't work with Pony, that it's probably SD1.5. As SDXL do typically work with Pony.
Does this work with non-anthro? (Anime mostly) and if not, Any plans for a non-anthro version?
Typically, Loras are pretty flexible. I've found that even Loras that are trained (probably) entirely on anime characters, can be made 3D, or even realistic. It's actually rather fascinating, I've even seen what real life Wilykat(Thundercats 2011) may look like, and real life Naruto. Although it may depend on the amount and quality of the training performed on the Lora.
Whats is Zoinkscoob? You say you reccomend it but looking it up on this site or on google, i dont really get a solid answer. I only see a checkpoint that is apparently a merge of it? Is that the one you mean? Thanks!
P.S: Also yes, i would love a pony, or ideally, an illustrious/NAI version of this as other have mentioned! Especially because, while other Vore LORAs exist, none to my knowledge have been trained on disposal or some other things. Thanks again!
Details
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.