Everyone's favorite short king.
This is model is the culmination of multiple Rocket Raccoon datasets I've prepared over the past while. Resulting in a what I feel is a great representation of the character that is not limited to one style or concept.
For the "formula", I used the following, along with their respective tags.
mcu r0cket - 142 images from the MCU movies.
e1dos r0cket - 56 images from the Eidos game.
s1dekick r0cket - 45 images from the artist Sidekick.
92 images from various other fanarts and nsfw art.
For the MCU, Eidos, and Sidekick datasets, I kept the same tags. So you should be able to skew your look towards one of those versions if you include those datasets' trigger words. I intentionally left out the Neelix dataset as I felt it was a bit too different. I do still recommend checking those versions out if you're only interested in those.
Description
Trained with the below params:
335 images:
{
"engine": "kohya",
"unetLR": 0.0005,
"clipSkip": 2,
"loraType": "lora",
"keepTokens": 1,
"networkDim": 16,
"numRepeats": 5,
"resolution": 1024,
"lrScheduler": "cosine",
"minSnrGamma": 5,
"noiseOffset": 0.1,
"targetSteps": 6282,
"enableBucket": true,
"networkAlpha": 8,
"optimizerType": "Prodigy",
"textEncoderLR": 0.00005,
"maxTrainEpochs": 15,
"shuffleCaption": true,
"trainBatchSize": 4,
"flipAugmentation": false,
"lrSchedulerNumCycles": 3
}
FAQ
Details
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.