VisualNote Portrait Illustration Styles DoRA All image labels and prompt guide can be seen at: https://www.seaart.ai/articleDetail/d737hr5e878c73dlddrg Key changes: - Removed convolution Dim and Alpha - Changed from adamW8Bit to Prodigy; Unet learning rate = 1, Text Encoder learning rate = 1. - From LoRa to DoRa - From 5000 total steps to 1500 steps. - Network Dim and Alpha from 128/64 to 64/32. Parameter Settings Network Module DoRA Use Base Model Nova Anime XL - IL v17.0 Trigger words VisualNote Image Processing Parameters Repeat 1 Epoch 10 Save Every N Epochs 1 Training Parameters Seed - Clip Skip 2 Text Encoder learning rate 1 Unet/DiT learning rate 1 LR Scheduler constant_with_warmup Optimizer Prodigy Network Dim 64 Network Alpha 32 Gradient Accumulation Steps - Label Parameters Shuffle caption - Keep n tokens - Advanced Parameters Noise offset 0.03 Multires noise discount 0.1 Multires noise iterations 10 conv_dim - conv_alpha - Batch Size 1 Sample Image Settings Prompt VisualNote, Digital anime-style drawing of Chun-Li from Street Fighter, kneeling with a determined expression, wearing her iconic blue qipao with gold accents, and brown tights. masterpiece, best quality, amazing quality, high quality, Sampler euler Steps: 1540 Loss: 0.0878