photo backgroundのタグに新たなバリエーションの追加を目指した。
効果: 背景の写実感UP。背景の強調。キャラが遠ざかる。
V1 is a strong influence.
V2 is a good influence.
Learned to add a new variation on the "photo background" tag.
The training data is realistic only for the background, reducing the negative impact on the anime character.
effect
・Increased realism background
・Background emphasis.characters move away a bit.
Description
illustrious based.The base model was changed to an illustrious prototype.
Thanks to this, the negative effects disappeared and the result was good.
Custom model had overlearning bias.
FAQ
Comments (2)
Great Lora! It is absolutely great. It provides very detailed backgrounds in such amazing camera angles. For me it worked best at 0.3. Great job!
Can you share the corresponding drawing prompts for some of the backgrounds in Lora?