Let me explain the difference between v1 and v2... Basically, the proportion of blurry and blurry background materials has increased. This protects the focus of learning for realistic portraits. Batch size 4 is not suitable for portraits where both portraits and landscapes are very clear; it's difficult to handle both simultaneously. If you give the materials to Gemini to draw anime characters or change anime faces (for realistic figures with petite figures), then Illustrious XL v0.1 can handle batch size 4. The proportion of figures in v2 has also increased significantly, so there's less need for detailed character descriptions, or the LoRa weighting has been reduced as before. Secondly, creator avoids blindly pursuing high image quality and high-definition retouching for large images. Many images on X, Pinterest, and Instagram now have high-definition retouching, so it's natural. PhotoScape X can batch process issues like oversaturated colors or details, or excessive noise. Personally, I use AI to remove some close-up clutter, such as zebra crossings or road patches, but this is time-consuming. (我解釋一下v1跟v2的差別...其實就是blurry跟blurry background的素材比例增加了，這對寫實人物的專注力學習是一種保護，batch size 4本來不適合寫實人物跟風景都很清楚，它很難同時兼顧兩者，如果把素材丟給gemini畫動漫人物或換動漫臉(身材要嬌小的寫實人物)，那Illustrious xl v0.1在batch size 4也應付得了。v2的人物比例也增加了很多，所以人物描述不用很多，或像以前一樣降低lora權重了。其次作者是不盲目追求大圖的高畫質跟高清修復，所以很自然。現在很多X, pinterest and instagram的圖都有高清修圖處理。PhotoScape X可以批量處理顏色或細節過度飽和或雜點過多的問題。我個人是會把一些近距離較雜亂的東西用ai消除，比如斑馬線或馬路補丁，但這比較花時間。)

Hello!
May I make a request?

I’d like to ask for a LoRA (Illustrious version) of a Japanese athletics stadium.
The reason is that prompts or existing LoRAs usually generate only foreign-style stadiums.

Thank you very much!