
This LoRA attempts to replicate paintings style.
Trigger words:
-
Global: "paify, fine art parody" (mandatory)
-
Baroque: "baroque"
-
Impressionism: "impressionism"
-
Realism: "realism"
-
Rococo: "rococo"
-
Romanticism: "romanticism"
Don't use Adetailer, it will erase the painting effect.
Favorite weights: 0.7(anime models) - 1(realistic models)
Especially for anime characters it can be used at very low heights for an overall effect.
At 0.5 weight:
At 0.2 weight:
Please note that the "Romanticism" tag is baked in the Stable Diffusion models, results will greatly vary according to the checkpoint.
Steps, Hires steps, CFG: doesn't matter
Hires Upscaler: 4x-UltraSharp
Checkpoints: works way better with realistic checkpoints, especially Dreamshaper, but is usable with all. If you have to use an anime checkpoint, I recommend semi-realistic ones such as DarkSun.
Impressionism style is the easiest to use with anime checkpoints.
Combining LoRAs: that's the main point, it is usable with realistic and anime character LoRAs.
Effects can be combined or used separately.
Trained with OneTrainer on public domain paintings provided by the National Gallery of Art.
描述:
训练词语: paify, fine art parody
名称: woafu_paintify.safetensors
大小 (KB): 18542
类型: Model
Pickle 扫描结果: Success
Pickle 扫描信息: No Pickle imports
病毒扫描结果: Success