MoVQ: Modulating Quantized Vectors for High-Fidelity Image Generation
This addresses artifacts in high-resolution image generation for computer vision applications, representing an incremental improvement over existing VQ models.
The paper tackles the problem of repeated artifacts in two-stage VQ generative models caused by quantization encoding similar patches identically, proposing modulation of quantized vectors and multichannel quantization to improve image quality. Results show greatly improved reconstructed image quality and high-fidelity generation on two benchmark datasets.
Although two-stage Vector Quantized (VQ) generative models allow for synthesizing high-fidelity and high-resolution images, their quantization operator encodes similar patches within an image into the same index, resulting in a repeated artifact for similar adjacent regions using existing decoder architectures. To address this issue, we propose to incorporate the spatially conditional normalization to modulate the quantized vectors so as to insert spatially variant information to the embedded index maps, encouraging the decoder to generate more photorealistic images. Moreover, we use multichannel quantization to increase the recombination capability of the discrete codes without increasing the cost of model and codebook. Additionally, to generate discrete tokens at the second stage, we adopt a Masked Generative Image Transformer (MaskGIT) to learn an underlying prior distribution in the compressed latent space, which is much faster than the conventional autoregressive model. Experiments on two benchmark datasets demonstrate that our proposed modulated VQGAN is able to greatly improve the reconstructed image quality as well as provide high-fidelity image generation.