ThinkDrag: Semantic Drag-Based Image Editing with Visual Reasoning
ThinkDrag treats point drags as semantic edit instructions, not just geometry.
The paper says sparse handle-and-target points can imply object- or scene-level changes that older drag editors often miss. ThinkDrag adds visual reasoning, including an optional trace that explains the intended edit before generation. It also introduces Drag Endpoint Markers to ground each drag in the image, plus a supervised semantic-drag dataset and DragBench++ benchmark. The authors report state-of-the-art results on generation quality and plausibility while keeping competitive point-following precision. HF Daily Papers' note
The paper says sparse handle-and-target points can imply object- or scene-level changes that older drag editors often miss. ThinkDrag adds visual reasoning, including an optional trace that explains the intended edit before generation. It also introduces Drag Endpoint Markers to ground each drag in the image, plus a supervised semantic-drag dataset and DragBench++ benchmark. The authors report state-of-the-art results on generation quality and plausibility while keeping competitive point-following precision. HF Daily Papers' note
score 4