Tag
MixTok 是一种混合视觉 tokenizer，将紧凑的 1D 全局表示与用于局部精修的稀疏 2D token 相结合，通过内容自适应路由器分配 token 预算，在 ImageNet 256x256 上以约 128/131 token 的平均预算达到 rFID 1.195 和 agFID 1.95 的竞争性重建与生成性能。
QuadTok introduces a hierarchical quadtree-based visual tokenizer that dynamically allocates tokens to intricate image regions, saving ~10% of tokens on ImageNet while enabling a 947M GPT-style model to reach 2.08 gFID on ImageNet 256x256 and zero-shot spatially controlled generation.
This paper proposes a unified rate-distortion perspective on discrete visual tokenization, resolving key questions about quantization objectives and comparisons, and shows that vector quantization achieves the lowest distortion under controlled conditions.
InsightTok introduces content-aware perceptual losses to improve discrete visual tokenization for better text and face reconstruction, enhancing autoregressive image generation quality.
This paper introduces DRoRAE, a method that improves visual tokenization by fusing multi-layer features from pretrained vision encoders rather than relying solely on the last layer. It demonstrates significant improvements in reconstruction and generation quality on ImageNet and establishes a scaling law between fusion capacity and performance.