This is a large-scale checkpoint model for generating photorealistic AI photos, which is the successor to the MoonMIX series of large models I released on Civitai. The new version, MoonFilm, is a fine-tuned large model trained using the expanded FilmGirl film style LoRA image dataset and Dreambooth training. The realism of its portraits is significantly improved; if you are seeking a truly realistic texture in AI photos, be sure to try out this large model!
Also recommended for use in combination with the TEXTUAL INVERSION models: bad_pictures and negative_hand Negative Embedding. These two Negative Embedding models can effectively correct and optimize images without significantly affecting their texture.
这是一个用于生成写实风格 AI 照片的 checkpoint 大模型，其前身是我在 C 站发布的 MoonMIX系列大模型。新版本MoonFilm是使用扩容后的FilmGirl胶片风 LoRA图像训练集，使用 Dreambooth 训练而成的微调大模型。其人像真实感显著提升，如果你非常追求 AI 照片的真实质感，请务必尝试下这个大模型！
同时推荐搭配TEXTUAL INVERSION模型：bad_pictures、negative_hand Negative Embedding。这两个Negative Embedding模型在不显著影响图像质感的同时，可以很好地起到图像修正优化效果。
