Search papers, labs, and topics across Lattice.
The paper introduces GIST, a training-free image compositor designed to harmonize visual elements from diverse sources in graphic design pipelines. GIST addresses the limitation of existing methods that assume input elements are already stylistically consistent, which is often not the case in real-world design scenarios. By integrating GIST with existing design methods like LaDeCo and Design-o-meter, the authors demonstrate significant improvements in visual harmony and aesthetic quality, as validated by LLaVA-OV and GPT-4V.
Mismatched visual elements torpedo design harmony, but GIST offers a training-free fix that stylistically blends components, boosting aesthetic quality in existing pipelines.
Graphic design creation involves harmoniously assembling multimodal components such as images, text, logos, and other visual assets collected from diverse sources, into a visually-appealing and cohesive design. Recent methods have largely focused on layout prediction or complementary element generation, while retaining input elements exactly, implicitly assuming that provided components are already stylistically harmonious. In practice, inputs often come from disparate sources and exhibit visual mismatch, making this assumption limiting. We argue that identity-preserving stylization and compositing of input elements is a critical missing ingredient for truly harmonized components-to-design pipelines. To this end, we propose GIST, a training-free, identity-preserving image compositor that sits between layout prediction and typography generation, and can be plugged into any existing components-to-design or design-refining pipeline without modification. We demonstrate this by integrating GIST with two substantially different existing methods, LaDeCo and Design-o-meter. GIST shows significant improvements in visual harmony and aesthetic quality across both pipelines, as validated by LLaVA-OV and GPT-4V on aspect-wise ratings and pairwise preference over naive pasting. Project Page: abhinav-mahajan10.github.io/GIST/.
CMU ML