MetaPoint: Unlocking Precise Spatial Control in Agentic Visual Generation
This work addresses the fundamental problem of precise spatial control in generative models for researchers and practitioners, offering a lightweight solution that leverages existing positional encodings.
MetaPoint introduces a method for precise spatial control in generative visual models by representing continuous 2D coordinates as special tokens, enabling pixel-level object positioning without architectural changes. The approach achieves compositional spatial primitives for agentic generation.
Generative visual models fundamentally struggle with precise spatial control. This arises from a core disconnect: models can process textual descriptions of space but cannot directly map numerical coordinates onto the 2D image canvas. We introduce MetaPoint, a method that bridges this gap by representing a continuous 2D coordinate as a single, special token. Crucially, MetaPoint requires no new architectural components; it directly leverages the model's inherent positional encoding schemes to interpret these coordinates, treating our token as a virtual point on the canvas. This lightweight approach enables pixel-level control of an object's position with one token or its bounding box with two, all without requiring architectural changes or bespoke attention masking. The MetaPoint tokens are designed to be compositional, serving as spatial primitives. This allows a planner agent to decompose a high-level user request into a structured sequence of primitives for the generator. By providing a simple, precise, and scalable building block for spatial control, MetaPoint unlocks more powerful compositional generative agents and enables intuitive, interactive editing systems.