This is my fourth style lora so go easy on me with this one. While I liked the style of V1, it actually looked nothing like the movie or the characters. Tried my hand again and managed to do a better job at it this time, but its still not perfect. I recommend starting with a weight of 0.8 for this one if the image you're going for is complex, then work your way up to 1 if you want a little more style to bleed in. Its trained on the 8 major characters in the movie. How well each character is trained depends on their screentime of course. Also remember, the movie doesn't have any nsfw so please don't expect too much from the model for that. The tags for the characters are below:Puss in Boots - puss, domestic cat, orange fur, green eyes, green sclera, hat, cape, bootsPerrito - p3rrito, chihuahua, blue eyes, brown fur, white fur, black furDeath - d3ath, wolf, male, red eyes, white fur, poncho, pants, arm wraps, leg wrapsKitty Softpaws - s0ftp4ws, domestic cat, black fur, white fur, blue eyesMama Bear - mb3ar, bear, brown fur, green eyes, pink beretPapa Bear - pb3ar, bear, brown fur, scarf, yellow eye, white eyeBaby Bear- bb3ar, bear, brown fur, blue eyesGoldilocks - g0ldilocks, human, blonde hair, twin buns, green eyes

Puss in Boots: The Last Wish Style + Characters

Cleaned up the dataset and retrained on higher step count. Would love to add more images to the dataset but I didn't want to watch the movie a third time lol. Training params are below. If I had to redo it, I might set repeats and noise to lower values. 315 images: { "engine": "kohya", "unetLR": 0.0005, "clipSkip": 2, "loraType": "lora", "keepTokens": 1, "networkDim": 32, "numRepeats": 4, "resolution": 1024, "lrScheduler": "cosine", "minSnrGamma": 5, "noiseOffset": 0.1, "targetSteps": 6300, "enableBucket": true, "networkAlpha": 16, "optimizerType": "Prodigy", "textEncoderLR": 0.00005, "maxTrainEpochs": 20, "shuffleCaption": true, "trainBatchSize": 4, "flipAugmentation": false, "lrSchedulerNumCycles": 3}

Files with SHA256: 40dd1c1703f120b672cab77d96cddcbeba536952b4badad21b5c5822816c55c3