UniCom: Unified Multimodal Modeling via Compressed Continuous Semantic Representations
This addresses a bottleneck in multimodal AI for tasks like image generation and editing, though it appears incremental as it builds on existing continuous representation methods.
The paper tackled the problem of suboptimal performance in unified multimodal models due to discretization or high-dimensional generative modeling by introducing UniCom, a framework using compressed continuous semantic representations, which achieved state-of-the-art generation performance and exceptional controllability in image editing.
Current unified multimodal models typically rely on discrete visual tokenizers to bridge the modality gap. However, discretization inevitably discards fine-grained semantic information, leading to suboptimal performance in visual understanding tasks. Conversely, directly modeling continuous semantic representations (e.g., CLIP, SigLIP) poses significant challenges in high-dimensional generative modeling, resulting in slow convergence and training instability. To resolve this dilemma, we introduce UniCom, a unified framework that harmonizes multimodal understanding and generation via compressed continuous representation. We empirically demonstrate that reducing channel dimension is significantly more effective than spatial downsampling for both reconstruction and generation. Accordingly, we design an attention-based semantic compressor to distill dense features into a compact unified representation. Furthermore, we validate that the transfusion architecture surpasses query-based designs in convergence and consistency. Experiments demonstrate that UniCom achieves state-of-the-art generation performance among unified models. Notably, by preserving rich semantic priors, it delivers exceptional controllability in image editing and maintains image consistency even without relying on VAE.