Verified: 2 years ago
SafeTensor
Type | |
Stats | 1,873 |
Reviews | (239) |
Published | Apr 3, 2023 |
Base Model | |
Training | Steps: 1,600 |
Trigger Words | pikapoke pikachu |
Hash | AutoV2 014636E716 |
Pikachu Lora trained model based on hyperrealistic reproduction.
Here's one of my first LORA models.
Based on SD1.5 and trained with 512*512 pictures. It can easily support upscaling with x4_ultrasharp for more crisps details.
add an extra "pokemon" in prompt for more context so your model can use it right. Mostly tested with RPGV4