This Repository for analysis VAR(Visual Autoregressive Modeling)

     

Visual Autoregressive Modeling: Scalable Image Generation via Next-Scale Prediction

Analysis

How Model Generate Image?

Generate Image with depth 30 model

• Model Generate Image Coarse-to-Fine
• low resolution token determines the overall color
• high resolution token gradually adds detailed information in a residual manner

Failure Image

• Model can't generate person (I think this is becuase there is no prior on people)
• Model can't generate when there are multiple objects (I guess...)

License

This project is licensed under the MIT License - see the LICENSE file for details.

Citation

If our work assists your research, feel free to give us a star ⭐ or cite us using:

@Article{VAR,
      title={Visual Autoregressive Modeling: Scalable Image Generation via Next-Scale Prediction}, 
      author={Keyu Tian and Yi Jiang and Zehuan Yuan and Bingyue Peng and Liwei Wang},
      year={2024},
      eprint={2404.02905},
      archivePrefix={arXiv},
      primaryClass={cs.CV}
}