Not loaded
Download
1 variant available
bf16 SafeTensor
H3_Ref2V_Anime_Slider_v1.safetensors
BF16, good balance • 73.99 MB
Verified: 20 hours ago
81
362
Generation, training and LoRA distribution on Civitai are covered by Civitai’s own license agreement with MiniMax. If you download these weights and run them yourself, your use is instead governed by the MiniMax H3 Community License Agreement, whose grant excludes the European Union, the United Kingdom, the Republic of Korea and the United States of America.
MiniMax H3
This is a slider LoRA designed to control the overall style, shifting between 2D anime and photorealistic aesthetics.
Positive values lean towards a 2D anime style, while negative values lean towards photorealism.
The recommended range is -5 to 5; typically, a setting of 3 or -3 is sufficient to define the style, though the strength may need adjustment depending on the specific prompts and reference images used.
The LoRA was originally created because the base H3 model's 2D anime style leaned too heavily towards 3D and lacked a sufficiently flat, 2D look.
Consequently, its primary function is to push the style closer to 2D anime; while it can also shift anime characters toward a realistic look, that is not its main capability.
It is suitable for both Ref2V (Reference-to-Video) and T2V (Text-to-Video) workflows.
這是一個滑塊Lora , 用來控制整體風格是2d動漫風格或是寫實風格
正強度偏向2d動漫風格，負強度則為寫實
建議是 -5 ~ 5 ,一般情況3 or -3 就能決定風格 ，但還須視提示詞與參考圖不同來調整強度
一開始訓練此lora的目的其實是因為H3的2d動漫風格比較偏向3d，不夠平面化
所以其主要功能是讓風格更趨向2d動漫 , 雖然也有讓動漫人物轉真實的效果但這不是其主要的能力
適用於ref2v 或是T2V