๐ ๐๐๐ฆ๐ ๐๐ ๐ฌ๐ข๐จ ๐๐๐๐:
๐ฎ.๐ฑ๐ ๐ฝ๐ฎ๐ถ๐ป๐๐ฒ๐ฟ๐น๐ ๐ฎ๐ป๐ถ๐บ๐ฒ ๐๐๐๐น๐ฒ ๐๐ถ๐๐ต ๐ฏ๐ฟ๐ถ๐ด๐ต๐ ๐๐ถ๐ป๐๐ฎ๐ด๐ฒ ๐ฐ๐ผ๐น๐ผ๐ฟ๐ & ๐ฑ๐๐ป๐ฎ๐บ๐ถ๐ฐ ๐น๐ถ๐ด๐ต๐๐ถ๐ป๐ด.
๐ฅ๐ฒ๐ฐ๐ผ๐บ๐บ๐ฒ๐ป๐ฑ๐ฒ๐ฑ ๐๐ฒ๐ถ๐ด๐ต๐: ๐ฌ.๐ฏ - ๐ญ
๐๐ป๐ถ๐บ๐ฎ ๐๐ฒ๐ฟ๐๐ถ๐ผ๐ป๐:
๐ฆ๐๐บ๐บ๐ฒ๐ฟ/๐ฆ๐ฝ๐ฟ๐ถ๐ป๐ด/๐ช๐ถ๐ป๐๐ฒ๐ฟ: All versions are results from the same training dataset, so thereโs not much difference between them. I tend to switch between versions until I get what I like. Basically, I name each version based on the lighting effect in the images using the same prompts: Summer has the brightest lighting, followed by Spring, then Winter.
(But I havenโt tested them much, so the names may not be that reliable lmao)
-----------------
๐๐ฟ๐ฒ๐ฎ๐ฎ ๐๐ฒ๐ฟ๐๐ถ๐ผ๐ป๐:
I like using it with semirealistic/realistic/Niji style base models & loras to add the softness and painterly feel to the images.
๐ฆ๐ผ๐ณ๐: It's better to use it at a high weight (around 1) & combine it with base models & loras that have a strong style.
๐ฆ๐ต๐ฎ๐ฑ๐ผ๐/๐ ๐ผ๐ผ๐ฑ/๐๐ฟ๐ถ๐ด๐ต๐: Strong effect (a bit overfitted sometime). It's better to use it at a weight of around 0.3 - 0.7 if you combine it with other LoRAs. All versions are results of the same training dataset so there's not much difference between them. I mainly use the Shadow ver.
It's better to use a canvas size higher than 1024 x 1024. My default canvas size: 1024 x 2048 (no need upscaling)
Description
E9
