Search papers, labs, and topics across Lattice.
This paper introduces Ideal, an In-depth Alignment framework that enhances discrete representation autoencoders (RAEs) by aligning quantized tokens with both shallow and deep features from pretrained vision foundation models (VFMs). The method addresses the issue of suboptimal reconstruction quality in RAEs, which often stems from the loss of fine-grained visual detail during discretization. Experimental results show that Ideal achieves a significant improvement in reconstruction performance, with a rFID score of 0.61 on ImageNet and sets a new state of the art for autoregressive image generation with a gFID of 1.89.
By leveraging the complementary strengths of shallow and deep VFM features, Ideal dramatically enhances image reconstruction quality and sets new benchmarks in autoregressive image generation.
Built on pretrained vision foundation models (VFMs), representation autoencoders (RAEs) have recently emerged as a promising approach for constructing semantically rich latent spaces for image generation. However, their reconstruction quality often remains suboptimal, largely because deep VFM representations do not preserve sufficient fine-grained visual detail. This limitation becomes even more severe after discretization, where missing low-level information is difficult to recover. In fact, we observe that shallow VFM features retain considerably richer local appearance and structural detail, which complements the high-level semantics carried by deep features used in existing RAEs. Motivated by this complementary property, we propose Ideal, an In-depth Alignment framework for discrete representation autoencoding. By jointly aligning quantized tokens with both shallow and deep VFM features, Ideal enables the resulting discrete visual tokens to preserve both visual fidelity and rich semantics. Extensive experiments demonstrate that Ideal yields superior reconstruction performance, achieving 0.61 rFID on ImageNet and outperforming the previous best method by 0.28. When used for autoregressive image generation, Ideal further produces a gFID of 1.89, establishing a new state of the art for autoregressive image generation.
