V2:34-image Lora model capable of generating images in the style of Furgeta.Prompts: by furgetaNegative prompts: text, speech bubble, dialogue, worst quality, low quality, blurry, artist name, watermark, username, signature, monochrome, greyscale, black and white, copyright symbol, sketch, 2 panel comic, 3 panel comic, 2koma, 3komaBase model: NoobAI-XL EPS 1.1Epochs: 20Repeats: 18Train Batch Size: 6Steps: 2040Resolution: 1024Enable Bucket: OnShuffle Tags: OnKeep Tokens: 2Clip Skip: 2Flip Augmentation: OnUnet LR: 0.0005Text Encoder LR: 0.00005LR Scheduler: cosine_with_restartsLR Scheduler Cycles: 3Min-SNR Gamma: 5Network Dim: 32Network Alpha: 16Noise Offset: 0.1Optimizer: AdamW8BitV1:25-image Lora model capable of generating images in the style of Furgeta.Prompts: by furgetaNegative prompts: text, speech bubble, dialogue, worst quality, low quality, blurry, artist name, watermark, username, signature, monochrome, greyscale, black and white, copyright symbol, sketch, 2 panel comic, 3 panel comic, 2koma, 3komaBase model: NoobAI-XL EPS 1.1Repeats: 24Steps: 2000epochs: 20Unet LR: 0.0001Text Encoder LR: 0.00001LR Scheduler: constantLR Scheduler Cycles: 1Min-SNR Gamma: 5Network Dim: 32Network Alpha: 16

Furgeta Style - NoobAI

An update where I decided to experiment with using Civitai’s default training settings, since the previous version were trained using settings similar to TensorArt. Let me know which version you like better, the one trained with Civitai’s default settings or the previous one.

Files with SHA256: 38391fd6b8718cca1db478820045165297538dcb0026214d3b8fd152af01af3b