ViTok-v2: Scaling Native Resolution Auto-Encoders to 5 Billion Parameters

• URL: http://arxiv.org/abs/2605.05331v1
• Date: Wed, 06 May 2026 18:03:13 GMT
• Title: ViTok-v2: Scaling Native Resolution Auto-Encoders to 5 Billion Parameters
• Abstract summary: Vision Transformer (ViT) autoencoders have emerged as compelling tokenizers for images, offering improved reconstruction over convolutional tokenizers.<n>ViTok found that the compression ratio r mediates a reconstruction-generation trade-off where lower r means better reconstructions but harder generations.<n>We introduce ViTok-v2, which addresses these limitations with native resolution support via NaFlex for generalization across resolutions and aspect ratios.
• Score: 30.67825027089147
• License: http://creativecommons.org/licenses/by-nc-nd/4.0/
• Abstract: Vision Transformer (ViT) autoencoders have emerged as compelling tokenizers for images, offering improved reconstruction over convolutional tokenizers. However, existing ViT tokenizers cannot explore this landscape as performance degrades outside training resolutions, and reliance on adversarial losses prevents stable scaling. ViTok (Hansen-Estruch et al., 2025) found that the compression ratio r mediates a reconstruction-generation trade-off where lower r means better reconstructions but harder generations, so improving tokenizer reconstruction is key to more Pareto-optimal tokenizers. We introduce ViTok-v2, which addresses these limitations with native resolution support via NaFlex for generalization across resolutions and aspect ratios, and a novel DINOv3 perceptual loss that replaces both LPIPS and GAN objectives for stable training at any scale. ViTok-v2 is trained on about 2B images and scaled to 5B parameters, the largest image autoencoder to date. ViTok-v2 matches or exceeds state-of-the-art reconstruction at 256p and outperforms all baselines at 512p and above. In joint scaling experiments with flow matching generators, we show that scaling both the autoencoder and the generator advances the Pareto frontier of this trade-off.

Related papers

• TC-AE: Unlocking Token Capacity for Deep Compression Autoencoders [51.71228803075235]
  We propose TC-AE, a ViT-based architecture for deep compression autoencoders.<n>We decompose token-to-latent compression into two stages, reducing structural information loss.<n>We enhance the semantic structure of image tokens via joint self-supervised training, leading to more generative-friendly latents.
  arXiv  Detail & Related papers  (2026-04-08T17:53:52Z)
• SFTok: Bridging the Performance Gap in Discrete Tokenizers [72.9996757048065]
  We propose textbfSFTok, a discrete tokenizer that incorporates a multi-step iterative mechanism for precise reconstruction.<n>At a high compression rate of only 64 tokens per image, SFTok achieves state-of-the-art reconstruction quality on ImageNet.
  arXiv  Detail & Related papers  (2025-12-18T18:59:04Z)
• Diffusion Transformers with Representation Autoencoders [35.43400861279246]
  Latent generative modeling, where a pretrained autoencoder maps pixels into a latent space for the diffusion process, has become the standard strategy for Diffusion Transformers (DiT)<n>Most DiTs continue to rely on the original VAE encoder, which introduces several limitations.<n>In this work, we explore replacing the VAE with pretrained representation encoders paired with trained decoders, forming what we term Representation Autoencoders (RAE)
  arXiv  Detail & Related papers  (2025-10-13T17:51:39Z)
• SSDD: Single-Step Diffusion Decoder for Efficient Image Tokenization [56.12853087022071]
  We introduce a new pixel diffusion decoder architecture for improved scaling and training stability.<n>We use distillation to replicate the performance of the diffusion decoder in an efficient single-step decoder.<n>This makes SSDD the first diffusion decoder optimized for single-step reconstruction trained without adversarial losses.
  arXiv  Detail & Related papers  (2025-10-06T15:57:31Z)
• GigaTok: Scaling Visual Tokenizers to 3 Billion Parameters for Autoregressive Image Generation [81.58846231702026]
  We introduce GigaTok, the first approach to improve image reconstruction, generation, and representation learning when scaling visual tokenizers.<n>We identify the growing complexity of latent space as the key factor behind the reconstruction vs. generation dilemma.<n>By scaling to $bf3 space billion$ parameters, GigaTok achieves state-of-the-art performance in reconstruction, downstream AR generation, and downstream AR representation quality.
  arXiv  Detail & Related papers  (2025-04-11T17:59:58Z)
• Learnings from Scaling Visual Tokenizers for Reconstruction and Generation [30.942443676393584]
  Visual tokenization via auto-encoding empowers state-of-the-art image and video generative models by compressing pixels into a latent space.<n>Our work aims to conduct an exploration of scaling in auto-encoders to fill in this blank.<n>We train ViTok on large-scale image and video datasets far exceeding ImageNet-1K, removing data constraints on tokenizer scaling.
  arXiv  Detail & Related papers  (2025-01-16T18:59:04Z)
• ConvLLaVA: Hierarchical Backbones as Visual Encoder for Large Multimodal Models [77.59651787115546]
  High-resolution Large Multimodal Models (LMMs) encounter the challenges of excessive visual tokens and quadratic visual complexity. We propose ConvLLaVA, which employs ConvNeXt, a hierarchical backbone, as the visual encoder of LMM. ConvLLaVA compresses high-resolution images into information-rich visual features, effectively preventing the generation of excessive visual tokens.
  arXiv  Detail & Related papers  (2024-05-24T17:34:15Z)
• ViTAR: Vision Transformer with Any Resolution [80.95324692984903]
  Vision Transformers experience a performance decline when processing resolutions different from those seen during training. We introduce fuzzy positional encoding in the Vision Transformer to provide consistent positional awareness across multiple resolutions. Our resulting model, ViTAR, demonstrates impressive adaptability, achieving 83.3% top-1 accuracy at a 1120x1120 resolution and 80.4% accuracy at a 4032x4032 resolution.
  arXiv  Detail & Related papers  (2024-03-27T08:53:13Z)
• ConvNeXt-ChARM: ConvNeXt-based Transform for Efficient Neural Image Compression [18.05997169440533]
  We propose ConvNeXt-ChARM, an efficient ConvNeXt-based transform coding framework, paired with a compute-efficient channel-wise auto-regressive auto-regressive. We show that ConvNeXt-ChARM brings consistent and significant BD-rate (PSNR) reductions estimated on average to 5.24% and 1.22% over the versatile video coding (VVC) reference encoder (VTM-18.0) and the state-of-the-art learned image compression method SwinT-ChARM.
  arXiv  Detail & Related papers  (2023-07-12T11:45:54Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.