This is one of my attempts. I previously trained several versions of RealUictic to Anime style LORA models. Later, a friend suggested I try Any Real to Anime, which led to this LORA model. Due to the limited test dataset, the style conversion is somewhat fixed, but the results are quite impressive.
trigger
real to any You can click the sample workflow below to experience it for free in the cloud.
workflow
This version introduces relatively minor changes to the partitioning
Description
This version significantly impacts the overall art style, altering the game's aesthetic.
FAQ
Comments (3)
How do these versions differ? Are they doing the same, or is there a difference in the final result?
In principle, they are the same, only the training methods differ. V2 will make more significant changes to the overall art style.
@shadowsii Got it, thank you! :)