CatalogStitch: Dimension-Aware and Occlusion-Preserving Object Compositing for Catalog Image Generation

arXiv cs.CV / 4/13/2026

📰 NewsSignals & Early TrendsIdeas & Deep AnalysisModels & Research

共有:

Key Points

CatalogStitch is presented as a set of model-agnostic techniques for dimension-aware and occlusion-preserving object compositing aimed at real-world catalog image generation.
It automates mask adaptation when product dimensions or aspect ratios differ, removing the need for users to manually adjust insertion regions.
It also introduces an occlusion-aware hybrid restoration approach to preserve occluding (foreground/background overlap) elements pixel-perfectly without requiring post-generation editing.
The paper adds CatalogStitch-Eval, a 58-example benchmark focused on aspect-ratio mismatch and occlusion-heavy catalog scenarios, along with PDF/HTML viewers for evaluation.
Experiments with three state-of-the-art compositing models (ObjectStitch, OmniPaint, InsertAnything) show consistent improvements, targeting reduced manual workflow effort in production settings.

Abstract

Generative object compositing methods have shown remarkable ability to seamlessly insert objects into scenes. However, when applied to real-world catalog image generation, these methods require tedious manual intervention: users must carefully adjust masks when product dimensions differ, and painstakingly restore occluded elements post-generation. We present CatalogStitch, a set of model-agnostic techniques that automate these corrections, enabling user-friendly content creation. Our dimension-aware mask computation algorithm automatically adapts the target region to accommodate products with different dimensions; users simply provide a product image and background, without manual mask adjustments. Our occlusion-aware hybrid restoration method guarantees pixel-perfect preservation of occluding elements, eliminating post-editing workflows. We additionally introduce CatalogStitch-Eval, a 58-example benchmark covering aspect-ratio mismatch and occlusion-heavy catalog scenarios, together with supplementary PDF and HTML viewers. We evaluate our techniques with three state-of-the-art compositing models (ObjectStitch, OmniPaint, and InsertAnything), demonstrating consistent improvements across diverse catalog scenarios. By reducing manual intervention and automating tedious corrections, our approach transforms generative compositing into a practical, human-friendly tool for production catalog workflows.