Awesome-Visual-Tokenizer

Categorize by Tokenizer Architecture:

• Image Tokenizers
  • Continuous 2D CNN Image Tokenizer
  • Discrete 2D CNN Image Tokenizer
  • 2D Transformer Image Tokenizer
  • 1D Image Tokenizer
  • Image Tokenizer with Generative Decoder
• Video Tokenizers
  • Continuous 3D CNN Video Tokenizer
  • Discrete 3D CNN Video Tokenizer
  • 3D Transformer Video Tokenizer
  • 1D Video Tokenizer
  • Video Tokenizer with Generative Decoder
• Others

Image Tokenizers

Continuous 2D CNN Image Tokenizer

• Diagnosing and enhancing VAE models

  • Dai, B., Wipf, D. et al.
  • arXiv:1903.05789
  • 2019

• Deep compression autoencoder for efficient high-resolution diffusion models

  • Chen, J., Cai, H., Chen, J. et al.
  • arXiv:2410.10733
  • 2024

• DC-AE 1.5: Accelerating Diffusion Model Convergence with Structured Latent Space

  • Chen, J., Zou, D., He, W. et al.
  • arXiv:2508.00413
  • 2025

Discrete 2D CNN Image Tokenizer

• Taming transformers for high-resolution image synthesis

  • Esser, P., Rombach, R., Ommer, B. et al.
  • CVPR 2021
  • 2021

• Autoregressive image generation using residual quantization

  • Lee, D., Kim, C., Kim, S. et al.
  • CVPR 2022
  • 2022

• Finite scalar quantization: Vq-vae made simple

  • Mentzer, F., Minnen, D., Agustsson, E. et al.
  • arXiv:2309.15505
  • 2023

• Language Model Beats Diffusion--Tokenizer is Key to Visual Generation

  • Yu, L., Lezama, J., Gundavarapu, N.B. et al.
  • arXiv:2310.05737
  • 2023

• Magvit: Masked generative video transformer

  • Yu, L., Cheng, Y., Sohn, K. et al.
  • CVPR 2023
  • 2023

• Emu3: Next-Token Prediction is All You Need

  • BAAI, E.T.
  • arXiv:2409.18869
  • 2024

• Visual autoregressive modeling: Scalable image generation via next-scale prediction

  • Tian, K., Jiang, Y., Yuan, Z. et al.
  • arXiv:2404.02905
  • 2024

• Autoregressive Model Beats Diffusion: Llama for Scalable Image Generation

  • Sun, P., Jiang, Y., Chen, S. et al.
  • arXiv:2406.06525
  • 2024

• Open-magvit2: An open-source project toward democratizing auto-regressive visual generation

  • Luo, Z., Shi, F., Ge, Y. et al.
  • arXiv:2409.04410
  • 2024

• Quantize-then-Rectify: Efficient VQ-VAE Training

  • Zhang, B., Rao, Q., Zheng, W. et al.
  • arXiv:2507.10547
  • 2025

• Image and video tokenization with binary spherical quantization

  • Zhao, Y., Xiong, Y., Kr"ahenb"uhl, P. et al.
  • arXiv:2406.07548
  • 2024

• End-to-end vision tokenizer tuning

  • Wang, W., Zhang, F., Cui, Y. et al.
  • arXiv:2505.10562
  • 2025

2D Transformer Image Tokenizer

• Vector-quantized Image Modeling with Improved VQGAN

  • Yu, J., Li, X., Koh, J. et al.
  • arXiv:2110.04627
  • 2022

• Vision Foundation Models as Effective Visual Tokenizers for Autoregressive Image Generation

  • Zheng, A., Wen, X., Zhang, X. et al.
  • arXiv:2507.08441
  • 2025

• Learnings from scaling visual tokenizers for reconstruction and generation

  • Hansen-Estruch, P., Yan, D., Chung, C. et al.
  • arXiv:2501.09755
  • 2025

• MANZANO: A Simple and Scalable Unified Multimodal Model with a Hybrid Vision Tokenizer

  • Li, Y., Qian, R., Pan, B. et al.
  • arXiv:2509.16197
  • 2025

• Unitok: A unified tokenizer for visual generation and understanding

  • Ma, C., Jiang, Y., Wu, J. et al.
  • arXiv:2502.20321
  • 2025

• UniLiP: Adapting CLIP for Unified Multimodal Understanding, Generation and Editing

  • Tang, H., Xie, C., Bao, X. et al.
  • arXiv:2507.23278
  • 2025

• Gigatok: Scaling visual tokenizers to 3 billion parameters for autoregressive image generation

  • Xiong, T., Liew, J.H., Huang, Z. et al.
  • ICCV 2025
  • 2025

• Latent Diffusion Model without Variational Autoencoder

  • Shi, M., Wang, H., Zheng, W. et al.
  • arXiv:2510.15301
  • 2025

• Diffusion Transformers with Representation Autoencoders

  • Zheng, B., Ma, N., Tong, S. et al.
  • arXiv:2510.11690
  • 2025

• Aligning Visual Foundation Encoders to Tokenizers for Diffusion Models

  • Chen, B., Bi, S., Tan, H. et al.
  • arXiv:2509.25162
  • 2025

1D Image Tokenizer

• An image is worth 32 tokens for reconstruction and generation

  • Yu, Q., Weber, M., Deng, X. et al.
  • arXiv:2406.07550
  • 2024

• Elastictok: Adaptive tokenization for image and video

  • Yan, W., Mnih, V., Faust, A. et al.
  • arXiv:2410.08368
  • 2024

• Latent Denoising Makes Good Visual Tokenizers

  • Yang, J., Li, T., Fan, L. et al.
  • arXiv:2507.15856
  • 2025

• Language-guided image tokenization for generation

  • Zha, K., Yu, L., Fathi, A. et al.
  • CVPR 2025
  • 2025

• Highly Compressed Tokenizer Can Generate Without Training

  • Beyer, L.L., Li, T., Chen, X. et al.
  • arXiv:2506.08257
  • 2025

• Masked autoencoders are effective tokenizers for diffusion models

  • Chen, H., Han, Y., Chen, F. et al.
  • ICML 2025
  • 2025

• Softvq-vae: Efficient 1-dimensional continuous tokenizer

  • Chen, H., Wang, Z., Li, X. et al.
  • CVPR 2025
  • 2025

• Adaptive length image tokenization via recurrent allocation

  • Duggal, S., Isola, P., Torralba, A. et al.
  • First Workshop on Scalable Optimization for Efficient and Adaptive Foundation Models
  • 2024

• Democratizing text-to-image masked generative models with compact text-aware one-dimensional tokens

  • Kim, D., He, J., Yu, Q. et al.
  • arXiv:2501.07730
  • 2025

• FlexTok: Resampling Images into 1D Token Sequences of Flexible Length

  • Bachmann, R., Allardice, J., Mizrahi, D. et al.
  • ICML 2025
  • 2025

• Gigatok: Scaling visual tokenizers to 3 billion parameters for autoregressive image generation

  • Xiong, T., Liew, J.H., Huang, Z. et al.
  • ICCV 2025
  • 2025

• DetailFlow: 1D Coarse-to-Fine Autoregressive Image Generation via Next-Detail Prediction

  • Liu, Y., Qu, L., Zhang, H. et al.
  • arXiv:2505.21473
  • 2025

• One-d-piece: Image tokenizer meets quality-controllable compression

  • Miwa, K., Sasaki, K., Arai, H. et al.
  • arXiv:2501.10064
  • 2025

• Image Tokenizer Needs Post-Training

  • Qiu, K., Li, X., Chen, H. et al.
  • arXiv:2509.12474
  • 2025

• AliTok: Towards Sequence Modeling Alignment between Tokenizer and Autoregressive Model

  • Wu, P., Zhu, K., Liu, Y. et al.
  • arXiv:2506.05289
  • 2025

Image Tokenizer with Generative Decoder

• Diffusion Autoencoders: Toward a Meaningful and Decodable Representation

  • Preechakul, N., Chatthee, N., Wizadwongsa, S., Suwajanakorn, S.
  • CVPR 2022
  • 2022

• epsilon-vae: Denoising as visual decoding

  • Zhao, L., Woo, S., Wan, Z. et al.
  • arXiv:2410.04081
  • 2024

• Diffusion autoencoders are scalable image tokenizers

  • Chen, Y., Girdhar, R., Wang, X. et al.
  • arXiv:2501.18593
  • 2025

• FlexTok: Resampling Images into 1D Token Sequences of Flexible Length

  • Bachmann, R., Allardice, J., Mizrahi, D. et al.
  • ICML 2025
  • 2025

• D-AR: Diffusion via Autoregressive Models

  • Gao, Z., Shou, M.Z.
  • arXiv:2505.23660
  • 2025

• Generative Multimodal Pretraining with Discrete Diffusion Timestep Tokens

  • Pan, K., Lin, W., Yue, Z. et al.
  • CVPR 2025
  • 2025

• Flow to the mode: Mode-seeking diffusion autoencoders for state-of-the-art image tokenization

  • Sargent, K., Hsu, K., Johnson, J. et al.
  • arXiv:2503.11056
  • 2025

• " Principal Components" Enable A New Language of Images

  • Wen, X., Zhao, B., Elezi, I. et al.
  • arXiv:2503.08685
  • 2025

Video Tokenizers

Continuous 3D CNN Video Tokenizer

• Video generation models as world simulators

  • Brooks, T., Peebles, B., Holmes, C. et al.
  • The SORA
  • 2024

• Cogvideox: Text-to-video diffusion models with an expert transformer

  • Yang, Z., Teng, J., Zheng, W. et al.
  • arXiv:2408.06072
  • 2024

• Open-sora plan: Open-source large video generation model

  • Lin, B., Ge, Y., Cheng, X. et al.
  • arXiv:2412.00131
  • 2024

• Hunyuanvideo: A systematic framework for large video generative models

  • Kong, W., Tian, Q., Zhang, Z. et al.
  • arXiv:2412.03603
  • 2024

• Wan: Open and advanced large-scale video generative models

  • Wan, T., Wang, A., Ai, B. et al.
  • arXiv:2503.20314
  • 2025

• Ltx-video: Realtime video latent diffusion

  • HaCohen, Y., Chiprut, N., Brazowski, B. et al.
  • arXiv:2501.00103
  • 2024

• Cosmos world foundation model platform for physical ai

  • Agarwal, N., Ali, A., Bala, M. et al.
  • arXiv:2501.03575
  • 2025

• LeanVAE: An Ultra-Efficient Reconstruction VAE for Video Diffusion Models

  • Cheng, Y., Yuan, F. et al.
  • arXiv:2503.14325
  • 2025

• Wf-vae: Enhancing video vae by wavelet-driven energy flow for latent video diffusion model

  • Li, Z., Lin, B., Ye, Y. et al.
  • CVPR 2025
  • 2025

• Step-video-t2v technical report: The practice, challenges, and future of video foundation model

  • Ma, G., Huang, H., Yan, K. et al.
  • arXiv:2502.10248
  • 2025

• Waver: Wave Your Way to Lifelike Video Generation

  • Zhang, Y., Yang, H., Zhang, Y. et al.
  • arXiv:2508.15761
  • 2025

• Vidtwin: Video vae with decoupled structure and dynamics

  • Wang, Y., Guo, J., Xie, X. et al.
  • CVPR 2025
  • 2025

Discrete 3D CNN Video Tokenizer

• Language Model Beats Diffusion--Tokenizer is Key to Visual Generation

  • Yu, L., Lezama, J., Gundavarapu, N.B. et al.
  • arXiv:2310.05737
  • 2023

• Videopoet: A large language model for zero-shot video generation

  • Kondratyuk, D., Yu, L., Gu, X. et al.
  • arXiv:2312.14125
  • 2023

• Loong: Generating minute-level long videos with autoregressive language models

  • Wang, Y., Xiong, T., Zhou, D. et al.
  • arXiv:2410.02757
  • 2024

• Vidtok: A versatile and open-source video tokenizer

  • Tang, A., He, T., Guo, J. et al.
  • arXiv:2412.13061
  • 2024

• Omnitokenizer: A joint image-video tokenizer for visual generation

  • Wang, J., Jiang, Y., Yuan, Z. et al.
  • arXiv:2406.09399
  • 2024

• Cosmos world foundation model platform for physical ai

  • Agarwal, N., Ali, A., Bala, M. et al.
  • arXiv:2501.03575
  • 2025

3D Transformer Video Tokenizer

• MAGI-1: Autoregressive Video Generation at Scale

  • Teng, H., Jia, H., Sun, L. et al.
  • arXiv:2505.13211
  • 2025

• AToken: A Unified Tokenizer for Vision

  • Lu, J., Song, L., Xu, M. et al.
  • arXiv:2509.14476
  • 2025

• Hi-VAE: Efficient Video Autoencoding with Global and Detailed Motion

  • Liu, H., Sun, W., Zhang, Q. et al.
  • arXiv:2506.07136
  • 2025

1D Video Tokenizer

• Larp: Tokenizing videos with a learned autoregressive generative prior

  • Wang, H., Suri, S., Ren, Y. et al.
  • arXiv:2410.21264
  • 2024

• Learning 1D causal visual representation with de-focus attention networks

  • Tao, C., Zhu, X., Su, S. et al.
  • NeurIPS 2025
  • 2024

Video Tokenizer with Generative Decoder

• Rethinking video tokenization: A conditioned diffusion-based approach

  • Yang, N., Li, P., Zhao, L. et al.
  • arXiv:2503.03708
  • 2025

• REGEN: Learning Compact Video Embedding with (Re-) Generative Decoder

  • Zhang, Y., Mai, L., Mahapatra, A. et al.
  • arXiv:2503.08665
  • 2025

• Hi-VAE: Efficient Video Autoencoding with Global and Detailed Motion

  • Liu, H., Sun, W., Zhang, Q. et al.
  • arXiv:2506.07136
  • 2025

Others

• Bridging Continuous and Discrete Tokens for Autoregressive Visual Generation
  • Wang, Y., Lin, Z., Teng, Y. et al.
  • ICCV2025
  • 2025