This is my nineteenth style lora so go easy with me on this one.Summary of the 2 versions. V1.0: Main version of the lora, trained on the full dataset for illustrious and noobai. Indigo: Trained for <a rel="ugc" href="https://civarchive.com/models/579632/indigo-furry-mix-xl">Indigo Furry Mix</a> on the 3 main characters. Check the version details for more info. This lora aims to replicate the style of Anhes' cell-shaded art. It's mainly used in their Lush Puppies comic (which is awesome btw check it out), but I included plenty of other stuff to flesh it out. I've been wanting to do it since I made the lora for Anhes' other style, but I held off til now as I knew it would take forever, and it did. Extremely happy with the outcome though. The main cast of the comics are all trained. I only tested the males, but the females should be there too. Obviously, characters with more panels will be better trained. They're listed below with the tags I recommend. well trained:<ul><li>lug, wolf, white fur, white body, yellow eyes, neck tuft, muscular, bara</li><li>theo, beagle, white fur, brown fur, floppy ears, yellow eyes, short stack</li><li>phan, beagle, floppy ears, white fur, brown fur, yellow eyes, muscular</li><li>bruce, rottweiler, black fur, brown fur, muscular</li><li>carlo, pit bull, grey fur, grey body, muscular</li><li>nobu, malamute, orange fur, white fur, chubby</li></ul>lesser trained but kinda might work:<ul><li>clarisse</li><li>lenora</li><li>rorie</li><li>thomas</li></ul>

Anhes Lush Puppies Style + Characters

Trained on the below params. 781 images (new record for me): { "engine": "kohya", "unetLR": 0.0005, "clipSkip": 2, "loraType": "lora", "keepTokens": 1, "networkDim": 32, "numRepeats": 2, "resolution": 1024, "lrScheduler": "cosine", "minSnrGamma": 5, "noiseOffset": 0.05, "targetSteps": 7810, "enableBucket": true, "networkAlpha": 16, "optimizerType": "Prodigy", "textEncoderLR": 0.00005, "maxTrainEpochs": 20, "shuffleCaption": true, "trainBatchSize": 4, "flipAugmentation": false, "lrSchedulerNumCycles": 3}

Files with SHA256: dedfe58d675ff6be182615892a62f03b0b00b5d9be6ed7c1f72e79c1ee7e126e