Character Introduction: https://civarchive.com/posts/11758171
In canon lore, he is a dominant and broody boy. But I don't own the porn, so you should lewd him up however you want!
Loose instruction for V1.1 both Pony and Illustrious Versions:
Works best around 0.5-0.7 LoRA Strength for the stereotypical wolfboy looks. Lower strength when working with multiple LoRAs.
Higher CFG (4-7) for simple scenes, including solo, portrait, and teasing poses.
Lower CFG (3-4) for complex scenes, including duo focus, sex, or with multiple people.
Sampler: Euler a or DPM++ 2M Karras. Though I find better success with different model checkpoints.
Breaks at Realism Checkpoints.
It is serviceable for now, but I recommend Hi-Res Fix if you have the means to do so.
Improving progress has been slow, but here we go.
Keywords (V1.1+) for both Pony and Illustrious:
Ak3ru, wolf boy, animal ears, wolf ears, wolf tail with white tip, amber eyes, black hair, animal ear fluff, bangsSecondary keywords:
penis, testicles, flaccid, foreskin, erection, anus, parted bang, fangs, perfect teeth, muscular, toned maleOther appropriate themes:
male focus, yaoi, twink, 1boy, 2boys, solo focus, nude, collar, glossy, kemonomimi, pet play, bondage, restrainedDataset: 62 images, tagged
Training Params are at the end of this model page.
Version History:
v1.0 Pony (outdated): Has a lot of glitch-y and burn-in effects
Keywords (partially outdated): Akeru, wolf boy, animal ears, wolf ears, wolf tail, amber eyes, black hair, animal ear fluff, bangs, white tail tip
This is a cry for help!
This is my first attempt at creating a character LoRA. I followed a few tutorials, but it's still not producing the correct results.
My original purpose is to make a simple LoRA to easily depict an athletic wolf boy with:
Amber eyes
Black hair
White tail tip
With just these three requirements, I hoped to generate him and my other OC fox boy (Sota) together in a scene more easily. However, based on the initial testing, I feel like this one is missing the mark. I must be doing something wrong and have high expectations!
If you are a character LoRA creator/guru and would like to help me, please feel free to reach out! Thank you!
Dataset: 62 images, tagged
Training Params (For V1.1E20 Pony):
{
"engine": "kohya",
"unetLR": 0.0005,
"clipSkip": 2,
"loraType": "lora",
"keepTokens": 3,
"networkDim": 32,
"numRepeats": 4,
"resolution": 1024,
"lrScheduler": "cosine",
"minSnrGamma": 0,
"noiseOffset": 0.1,
"targetSteps": 1260,
"enableBucket": true,
"networkAlpha": 32,
"optimizerType": "Adafactor",
"textEncoderLR": 0.00005,
"maxTrainEpochs": 20,
"shuffleCaption": true,
"trainBatchSize": 4,
"flipAugmentation": false,
"lrSchedulerNumCycles": 3
}Training Params (For V1.1E20 Illustrious):
{
"engine": "kohya",
"unetLR": 0.0005,
"clipSkip": 2,
"loraType": "lora",
"keepTokens": 3,
"networkDim": 32,
"numRepeats": 4,
"resolution": 1024,
"lrScheduler": "cosine_with_restarts",
"minSnrGamma": 5,
"noiseOffset": 0.1,
"targetSteps": 1260,
"enableBucket": true,
"networkAlpha": 16,
"optimizerType": "Adafactor",
"textEncoderLR": 0.00005,
"maxTrainEpochs": 20,
"shuffleCaption": true,
"trainBatchSize": 4,
"flipAugmentation": false,
"lrSchedulerNumCycles": 3
}Description
Double Cope: It's better than the previous Pony version...
New data set, new tagging method, 110 images total.
Since most images were genned with Sym-Pony, You should use lower LoRA strength (<0.5) when generate with Sym-pony, Prefect Pony or WAI-PonyXL.
Again, it's better... Overdosing
Training Params:
{
"engine": "kohya",
"unetLR": 0.0005,
"clipSkip": 2,
"loraType": "lora",
"keepTokens": 3,
"networkDim": 32,
"numRepeats": 5,
"resolution": 1024,
"lrScheduler": "cosine",
"minSnrGamma": 0,
"noiseOffset": 0.03,
"targetSteps": 2200,
"enableBucket": true,
"networkAlpha": 32,
"optimizerType": "Prodigy",
"textEncoderLR": 0.00005,
"maxTrainEpochs": 20,
"shuffleCaption": true,
"trainBatchSize": 5,
"flipAugmentation": false,
"lrSchedulerNumCycles": 3
}
FAQ
Looks like we don't have an active mirror for this file right now.
CivArchive is a community-maintained index — we catalog mirrors that volunteers upload to HuggingFace, torrents, and other public hosts. Looks like no one has uploaded a copy of this file yet.
Some files do get recovered over time through contributions. If you're looking for this one, feel free to ask in Discord, or help preserve it if you have a copy.