This dataset comprises 8060 original images. The majority (95%) are unfiltered photos taken during a one-week trip. An additional 1% consists of carefully selected high-quality photos, 2% are my own drawings and paintings, and the remaining 2% are public domain images. The dataset was used to train a custom-designed diffusion model with a resolution of 512x512 on an NVidia 4090 GPU for a period of 4 days training from scratch.
所有圖片資料集包含 8060 張原始圖片。其中大部分（95%）是在一週旅行期間拍攝的未經塞選的照片。另外 1% 是挑選的高品質照片，2% 是我自己過去的繪畫與素描作品，其餘 2% 則是公共領域圖片。這個是一個重新設計與擴散模型，生成解析度為 512x512，從零開始在 4090 GPU 上訓練了 4 天。
