[Checkpoint] AlbedoBase XL
模型简介：AlbedoBase XL（SFW 和 NSFW）
不需要精炼机，含有VAE。
将负面提示留空通常可带来最佳质量。
~~~~~~~~~~~~~~~~~~~~~~~o(*￣︶￣*)o~~~~~~~~~~~~~~~~~~~~~~~
目标
Stable Diffusion XL 拥有 66 亿个参数，比 SD v1.5 版本大约增加了 6.6 倍，我相信这不只是一个数字，而是一个可以带来性能显著提升的数字。
我们已经意识到 SD v1.5 的整体性能已经提升到超乎想象的程度，这要归功于我们社区的大量贡献。因此，我正在努力完成这个 AlbedoBase XL 模型，以便在这个 XL 版本中也能最佳地重现 v1.5 中的性能改进。
我的目标是直接测试所有公开上传到 Civitai 的 Checkpoint 和 LoRA 的性能，并仅合并经过多次筛选后被判定为最佳的资源。这将超越 Midjourney 等公司的图像生成 AI 的性能。
截至目前，AlbedoBase XL v1.3 已合并 141 个选定检查点和 251 个 LoRA。
~~~~~~~~~~~~~~~~~~~~~~~o(*￣︶￣*)o~~~~~~~~~~~~~~~~~~~~~~~
日志
v2.1
重新合并并使用新的合并算法和公式将 v0.1 调整为 2.0。
~~~~~~~~~~~~~~~~~~~~~~~o(*￣︶￣*)o~~~~~~~~~~~~~~~~~~~~~~~
测试提示词和参数：
Prompt: Envision a scene that transcends the boundaries of photography, where a woman's deep melancholy is as tangible as the space she occupies. She is adrift inside a space station, her presence a poignant contrast against the mechanical backdrop. Her gaze, lost in the vastness of space, speaks volumes of unspoken stories. Focus intently on her hands - pale, almost ethereal, they are captured in an emotive gesture, perhaps tracing the cold glass of the window or floating gracefully in zero gravity. These hands are not just part of her; they are storytellers in their own right. The image should be infused with the distinct qualities of a shot taken on a Canon 5D Mark IV with Kodak Ektar film - the colors are deep and resonant, the details are sharp yet possess an organic quality, and there's a subtle grain that adds a layer of depth and authenticity. This isn't just a photograph; it's a moment of raw emotion, frozen in time, inviting the viewer to not just see, but to feel.
Steps: 60, Size: 720x1280, Seed: 9, Model: albedobaseXL_v21, Version: v1.7.0, Sampler: DPM++ 2M SDE Heun Karras, CFG scale: 7.5, Model hash: 1718b5bb2d
页:
[1]