The aim of this LoRA is to essentially convert your favorite 2d furry checkpoint into a realism checkpoint without losing too much of what makes the original special. For that reason I recommend you use it almost as a slider. I used a mix of my favorite checkpoints in the sample pics above to give you an idea of what I mean as a way of showing what it looks like on different models.
I wish I can say V2 is a flat upgrade over V1, but they both have there perks. Here's a summary. Feel free to test and choose what works for you. For both I still highly recommend using a face detailer.
v1.0: Intentionally overtrained on a small dataset (mostly wolves and tigers) to maximize realism, but at the cost of some usability. Basically high risk, high reward. Use at 0.6 weight. Higher weights can be done but at your own risk.
v2.0: Doubled the dataset size to include more species, also trained on more reasonable settings. This is the safer option but isn't as realistic as v1.0 as a result. Use-able up to 1.0 strength and beyond.
v1.0 Anima: First attempt at an Anima version. Anima obviously isn't meant to be used for realistic images so don't expect pure realism just yet.
The reason the lora has "Delken" in the name is because that's the AI director who's images predominantly make up the dataset. I would love to have a more varied dataset but I haven't found anyone who consistently matches the quality Delken in pure realism IMO. As of v2.0, I've started adding my own Chatgpt generated images, but I'm still more than open to suggestions.
Delken if you're reading this, thanks for the awesome pics and sorry that I stole them. Feel free to DM me if you want the name taken off the lora or for anything else.
Description
Trained on 146 images:
{
"engine": "kohya",
"unetLR": 0.0005,
"clipSkip": 2,
"loraType": "lora",
"keepTokens": 1,
"networkDim": 16,
"numRepeats": 7,
"resolution": 1024,
"lrScheduler": "cosine",
"minSnrGamma": 5,
"noiseOffset": 0.1,
"targetSteps": 5110,
"enableBucket": true,
"networkAlpha": 8,
"optimizerType": "Prodigy",
"textEncoderLR": 0.00005,
"maxTrainEpochs": 20,
"shuffleCaption": true,
"trainBatchSize": 4,
"flipAugmentation": true,
"lrSchedulerNumCycles": 3
}
FAQ
Comments (5)
As mentioned in the description, I'd love to make a V2 of this one day with a larger dataset, but I'm struggling to find consistently high quality, photorealistic furries. If you know of any artists, directors, etc please let me know I'd be super appreciative. Ideally they are like Delken who uses means other than illustrious and thus avoids that muddy look on their images.
I have a pretty solid realism dataset you can use if you want. Problem is that it kind of has an almost 3d-cg look to it still even to know I think its pretty dang good. So it really depends on how realistic it needs to be. The detail level on fur and scales is next level though.
This image I recently posted "https://civitai.com/images/124534989" is a decent example of the level of detail the images have, although the proportions and style in this specific image aren't exact because I was playing around with styles.
I personally would love to see either a separate version for scales and skin type species, or a version of this that does not force only fur.
@willywizard That's nice. what model did you use on the image?
@Crft Tbh I used 2 models. The base was an experimental model I’m working on that I haven’t released, and a really low denoise upscale pass with ZIT.