
Introducing Harvest, an anime lora style that was created with the intent to imitate a similar style to the Harvest Moon videogame series from the 2000s. :)
Important Notes:
-
Works best with illustration models but can work with more semi-realistic anime models.
-
The weight will change depending on the model used. (0.8-0.9 for models like Aingdiffusion. 0.6-0.7 for models like MixProV4)
-
Works best with anime girls, but can do landscapes, animals and vehicles. For other concepts, place more emphasis using brackets. Ex. (honda car:1.2)
-
Compatible with NSFW content.
-
Use the ADetailer SD extension to clean up the face as needed!
Additional Notes:
-
V2 model: Pivoted towards using a different training method. Drastically decreased the file size. Hugely improved the effect and clarity of the model to better imitate the desired style.
-
V1 model: Changed the formula. Increased the size and diversity of the dataset. Slightly increased color saturation. Dramatically scaled up the effect of this model.
-
Beta model: This is the beta version, which means it still has a long way to go. I'm releasing it just to see what it can do.
描述:
训练词语:
名称: HarvestV2.safetensors
大小 (KB): 21704
类型: Model
Pickle 扫描结果: Success
Pickle 扫描信息: No Pickle imports
病毒扫描结果: Success