发表机构
Computer Vision Center; Universitat Autònoma de Barcelona; Adobe Research（计算机视觉中心; 巴塞罗那自治大学; Adobe 研究院）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
UniSlider提出感知均匀滑块，通过LoRA训练与推理时重映射，使编辑强度线性对应感知变化，在300个连续编辑基准上超越现有方法。
AI 中文摘要
滑块为连续图像编辑提供了直观的界面。然而，在当前生成方法中，滑块仅仅是方法强度参数（如适配器系数、提示权重或插值因子）的重新缩放。这种强度与感知变化的关系不佳。随着滑块移动，图像可能部分回退，范围的长段没有可见差异，而短间隔则使图像突然变换。重新映射强度可以解决这种不均匀的节奏，但前提是轨迹是单调的，而当前方法并未强制这一点。因此，我们将滑块与强度区分开来，并要求从输入到输出的感知距离随滑块值线性增长。我们引入了UniSlider，一种轻量级LoRA，在几步编辑骨干上训练，使其强度接近这种理想滑块。几步采样使我们能够在像素空间中施加这一目标，而无需中间真实值，并且骨干网络的输出在全强度下得以保留。然而，低秩适配器无法使强度完全均匀。因此，我们的滑块是推理时对强度的重新映射，通过自适应采样获得。由于训练优化使轨迹单调，这种重新映射无需额外训练或参数即可弥合剩余差距。在一个包含300个连续编辑的新基准上，评估均匀性、单调性、编辑保真度和身份保持，UniSlider在所有先前方法中表现优异，并在用户研究中受到青睐。
英文摘要
Sliders provide an intuitive interface for continuous image editing. In current generative approaches, however, the slider is simply a rescaling of the method's strength parameter, such as an adapter coefficient, a prompt weight, or an interpolation factor. This strength relates poorly to perceptual change. The image can partially revert as the slider moves, long stretches of the range produce no visible difference, and short intervals transform the image abruptly. Remapping the strength could fix this uneven pace, but only if the trajectory is monotone, which current methods do not enforce. We therefore distinguish the slider from the strength, and require perceptual distance from the input to grow linearly with the slider value. We introduce UniSlider, a lightweight LoRA trained on a few-step editing backbone so that its strength approximates this ideal slider. Few-step sampling lets us impose this objective in pixel space without intermediate ground truth, and the backbone's output is preserved at full strength. However, a low-rank adapter cannot make the strength fully uniform. Our slider is thus an inference-time remapping of the strength, obtained by adaptive sampling. Since training optmizes to make the trajectory monotone, this remapping closes the remaining gap without extra training or parameters. On a new benchmark of 300 continuous edits evaluating uniformity, monotonicity, edit fidelity, and identity preservation, UniSlider outperforms all prior methods and is preferred in a user study.
CommentsProject page: https://color.cvc.uab.cat/unislider
