CHIMERA: Adaptive Cache Injection and Semantic Anchor Prompting for Zero-shot Image Morphing with Morphing-oriented Metrics

2025-12-07

AI Summary

Diffusion models for image morphing suffer from insufficient smoothness and semantic consistency, often yielding abrupt transitions or oversaturated artifacts. To address this, we propose a zero-shot image morphing framework: it caches multi-level features—downsampling, middle-layer, and upsampling blocks—via the DDIM inversion process, and introduces an adaptive cache injection mechanism to preserve structural integrity. Additionally, cross-domain shared semantic anchor prompts, generated by CLIP, guide the denoising process toward semantic alignment. We further design a novel global-local consistency scoring metric, enabling, for the first time, joint structural and semantic alignment without any training. Experiments on multiple benchmarks and user studies demonstrate significant improvements in morphing speed, visual naturalness, and semantic coherence, achieving state-of-the-art performance.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Machine Learning: Large Multimodal Models (LMMs)
• Natural Language Processing: Lexical Semantics and Morphology

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• User Modeling, Personalization and Recommendation: Accountability, Transparency, and Ethics for personalization
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

Diffusion models exhibit remarkable generative ability, yet achieving smooth and semantically consistent image morphing remains a challenge. Existing approaches often yield abrupt transitions or over-saturated appearances due to the lack of adaptive structural and semantic alignments. We propose CHIMERA, a zero-shot diffusion-based framework that formulates morphing as a cached inversion-guided denoising process. To handle large semantic and appearance disparities, we propose Adaptive Cache Injection and Semantic Anchor Prompting. Adaptive Cache Injection (ACI) caches down, mid, and up blocks features from both inputs during DDIM inversion and re-injects them adaptively during denoising, enabling spatial and semantic alignment in depth- and time-adaptive manners and enabling natural feature fusion and smooth transitions. Semantic Anchor Prompting (SAP) leverages a vision-language model to generate a shared anchor prompt that serves as a semantic anchor, bridging dissimilar inputs and guiding the denoising process toward coherent results. Finally, we introduce the Global-Local Consistency Score (GLCS), a morphing-oriented metric that simultaneously evaluates the global harmonization of the two inputs and the smoothness of the local morphing transition. Extensive experiments and user studies show that CHIMERA achieves smoother and more semantically aligned transitions than existing methods, establishing a new state of the art in image morphing. The code and project page will be publicly released.

Problem

• Achieving smooth, semantically consistent image morphing with diffusion models
• Addressing abrupt transitions and over-saturated appearances in morphing
• Evaluating global harmonization and local smoothness in morphing transitions

Innovation

• Adaptive Cache Injection for feature alignment
• Semantic Anchor Prompting for semantic bridging
• Global-Local Consistency Score for morphing evaluation