this was an experiment to see if i could guide the position of characters with a caption. it doesn't really work lol. also makes everything look ugly. u probably shouldn't use this on civitai or u'll lose a lot of buzz for nothing lmao
but if u gen dozens and dozens of times sometimes it at least gets the positions right. u can probably fix the artifacts by doing img2img without the lora or with the lora at a weaker strength or inpainting. all of these things are easier to do locally
v1
trigger: boy_character caught crossdressing by girl_character, sandwiched between left_character and right_character, bra, skirt, indoors, holding panties, clenching hand, 1girl, 3boys
Description
trained on 8 images
{
"engine": "kohya",
"unetLR": 0.0001,
"clipSkip": 1,
"loraType": "lora",
"keepTokens": 2,
"networkDim": 32,
"numRepeats": 15,
"resolution": 1024,
"lrScheduler": "cosine_with_restarts",
"minSnrGamma": 5,
"noiseOffset": 0,
"targetSteps": 1200,
"enableBucket": true,
"networkAlpha": 32,
"optimizerType": "Adafactor",
"textEncoderLR": 0.00005,
"maxTrainEpochs": 40,
"shuffleCaption": true,
"trainBatchSize": 4,
"flipAugmentation": false,
"lrSchedulerNumCycles": 4
}Details
Files
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.