Omni-3DEdit：单次前向的通用多能 3D 编辑
Omni-3DEdit: Generalized Versatile 3D Editing in One-Pass
把指令驱动的 3D 编辑从「2D 模型引导 + 逐场景迭代优化」改为单次前向的生成模型：沿序列维拼接源视图 latent 与条件 token，并用双流 LoRA 分离不同视图线索，可直接复用到任何需要多视图一致编辑或免优化 3D 编辑的场景。
原文依据 it can complete various 3D editing tasks in one forward pass, reducing the inference time from tens of minutes to approximately two minutes
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论大多数指令驱动的 3D 编辑方法依赖 2D 模型来引导对 3D 表示的显式迭代优化。然而这一范式存在两个主要缺陷。第一，它缺乏面向不同 3D 编辑任务的通用设计，因为对 3D 几何的显式操作需要依赖具体任务的规则，例如 3D 外观编辑要求保留源 3D 几何，而 3D 物体移除则会改变源几何。第二，迭代优化过程极为耗时，往往需要数千次调用 2D/3D 更新。我们提出 Omni-3DEdit，这是一个统一的、基于学习的模型，能够以隐式方式泛化到多种 3D 编辑任务。实现该目标的一个关键挑战是缺少成对的源—编辑后多视图素材用于训练。为解决这一问题，我们构建了一条数据流水线，合成了数量相对丰富的高质量成对多视图编辑样本。随后，我们将预训练生成模型 SEVA 作为骨干网络，通过在序列空间中拼接源视图 latent 与条件 token 对其进行适配。我们提出双流低秩适配模块来解耦不同的视图线索，大幅增强了模型的表征学习能力。作为一个基于学习的模型，我们的模型无需耗时的在线优化，可以在一次前向传播中完成多种 3D 编辑任务，将推理时间从数十分钟缩短至约两分钟。大量实验证明了 Omni-3DEdit 的有效性与高效性。
英文原文摘要 1559 字符
Most instruction-driven 3D editing methods rely on 2D models to guide the explicit and iterative optimization of 3D representations. This paradigm, however, suffers from two primary drawbacks. First, it lacks a universal design of different 3D editing tasks because the explicit manipulation of 3D geometry necessitates task-dependent rules, e.g., 3D appearance editing demands inherent source 3D geometry, while 3D removal alters source geometry. Second, the iterative optimization process is highly time-consuming, often requiring thousands of invocations of 2D/3D updating. We present Omni-3DEdit, a unified, learning-based model that generalizes various 3D editing tasks implicitly. One key challenge to achieve our goal is the scarcity of paired source-edited multi-view assets for training. To address this issue, we construct a data pipeline, synthesizing a relatively rich number of high-quality paired multi-view editing samples. Subsequently, we adapt the pre-trained generative model SEVA as our backbone by concatenating source view latents along with conditional tokens in sequence space. A dual-stream LoRA module is proposed to disentangle different view cues, largely enhancing our model's representational learning capability. As a learning-based model, our model is free of the time-consuming online optimization, and it can complete various 3D editing tasks in one forward pass, reducing the inference time from tens of minutes to approximately two minutes. Extensive experiments demonstrate the effectiveness and efficiency of Omni-3DEdit.