Setting the weights to approximately 1.0 is recommended.
If the generated image looks strange, set the CFG Scale to between 5 and 7.
Please Use NoobAI V-Pred 1.0 Checkpoint!
Activation tag : higurashikira, anime coloring, anime screencap
Recommend Prompt
Positive: higurashikira, anime screencap, anime coloring, (tag), (masterpiece, best quality, absurdres, highres, newest, very awa, year 2024:1.2), Negative: (worst quality, worst aesthetic, bad quality, old, early, mammal, anthro, furry, feral, semi-anthro, lowres, (bad), off-topic, multiple views, comic, very displeasing, displeasing, chromatic aberration, extra digit, fewer digits, fewer, text, error, artistic error, missing, jpeg artifacts, artist name, signature, twitter username, username, watermark, scan, unfinished, [abstract], variations, anatomical nonsense), If certain characters are not compatible with older 1.0 versions, it's advisable to use ControlNet.
Based on the data used for training SDXL v1.0, I've developed a Pony-exclusive LoRA!
This means it's not a LoRA trained with AnimagineXL-V3 as before, but one exclusively for Pony Diffusion.
Furude Rika = 1girl,solo,furude rika,blue hair,purple eyes,long hair,blunt bangs,bangs