Vore | Concept enhancer
This LoRA is a bit different from the usual type that I make, it aims to not particularly provide a concept on it's own, but rather to strenghten the consistency of other existing LoRAs.
How to use: To use this LoRA, simply add it to your generation along with other LoRAs that you wish, such as this one, for example. It does not require any activation tag, though you should make sure you to include tags such as "vore, oral vore, ..." to hone in on a specific concept.
Tags to be used include, but not limited to: "vore, oral vore, unbirthing, partially inside, anal vore, internal, cock vore, fully inside, ...", to be used along other LoRAs
This is still experimental, and I hope to make it way better in the future! For comparison, here are two sets of images both with and without the LoRA activated:
Set 1
Model: Artofura v2
LoRA: Vore | Feet first (v1.0) [Strenght:1.00]
Clip: 2
Steps: 22
CFG: 4.5
Prompt: vore_feet_first, snake, green scales, feral, feral pred, duo, human, human prey, male prey, nude, brown hair, mature human, forest, masterpiece, best quality, high quality, absurdres, highres, incredibly absurdres, newest, amazing quality, vore, swallowing, partially inside
Negative prompt: fewer digits, lowres, worst quality, bad quality, bad anatomy, sketch, jpeg artifacts, signature, watermark, old, oldest, censored, bar censor, headband, easynegative, watermark, patreon logo
Image width and height: 1152 x 896
Sampler: Euler A
Scheduler: Simple
Seed: 1234567890
Set 2
Model: waiNSFWIllustrious_v140
LoRA: Vore | Tongue wrap [Strenght:0.85]
Clip: 2
Steps: 22
CFG: 4.5
Prompt: vore_tongue_wrap, pokemon, lucario, bedroom, on bed, lying, anthro, anthro pred, side view, human, brown hair, human prey, detailed background, masterpiece, best quality, high quality, absurdres, highres, incredibly absurdres, newest, amazing quality, vore, oral vore, swallowing
Negative prompt: fewer digits, lowres, worst quality, bad quality, bad anatomy, sketch, jpeg artifacts, signature, watermark, old, oldest, censored, bar censor, headband, easynegative, patreon logo
Image width and height: 1152 x 896
Sampler: Euler A
Scheduler: Simple
Seed: 755883396
Description
FAQ
Comments (7)
This looks extremely promising and a refreshing drop of holy water in vore content.
Please remember to take breaks and avoid burnouts. Take your time because this looks like a really damn good idea and work.
Thank you VERY much for all your work thus far! <3
Thank you for the feedback!
Whilst this is an early version, I still hope to make this work even better in the future :D
Finally, now I can make my vore arts not some crappy mess!~ ^w^
Doesn't work with the feet first lora, it keeps inverting the pred and the prey
This issue is quite deep rooted, and I've yet to find a way to completely overcome it. The way I remedy it normally is by adding artificial tags to further indicate pred/prey when training (though this is not perfect and I still look for better ways to do it), but that's not something I've done for this version of this LoRA yet, as it has a way bigger dataset and it was initially a proof of concept. Hopefully this works better in the next version.
The only way to try and get around this 'stubbornness' from the models is to try different CFGs, adding strenght to the pred's tags and/or adding other descriptors to hone in the image, all of which are not ideal and I wish that the prompts could easily define roles.
Natural language models have a high advantage in this regard, hopefully in the future we'll be able to leverage that too.
Did you try inverting it in the prompt so it would be correct on the image than? (Sorry I just had to...)
Is this mostly use for things such as swallowing, big bellies and the likes? What about post vore stuff like disposal? Can it help with that?