Ethereum co-founder, Vitalik Buterin, recently expressed his support for the new image compression method called Token for Image Tokenizer (TiTok AI). Buterin believes that TiTok AI has the capability to compress images to a size that would allow them to be added to the blockchain. He explained on his social media account that the method could encode a profile picture to 320 bits, making it small enough to be stored on the blockchain for every user.
Buterin’s interest in TiTok AI was sparked by a post by a researcher from Leonardo AI, an artificial intelligence image generator platform. The researcher, known as @Ethan_smith_20, demonstrated how the method could encode complex images into 32 tokens, making it easier for developers and creators to produce profile pictures and non-fungible tokens (NFTs).
TiTok AI, a collaboration between TikTok’s parent company ByteDance and the University of Munich, offers a one-dimensional tokenization framework that differs from traditional two-dimensional methods. The AI technology behind TiTok enables it to compress 256 by 256-pixel images into 32 distinct tokens, addressing issues encountered with previous methods like VQGAN.
Previous image tokenization methods were limited by 2D grids and downsampling factors, resulting in difficulties handling redundancies and similarities in close regions of images. TiTok’s AI-driven approach solves these challenges by tokenizing images into 1D sequences, providing a compact representation and eliminating region redundancy.
In addition to improving image storage on blockchain platforms, TiTok offers significant enhancements in processing speed, boasting speeds up to 410 times faster than current technologies. This advancement in computational efficiency is a major breakthrough in the field of image tokenization.
Overall, Vitalik Buterin’s endorsement of TiTok AI highlights the potential for innovative solutions to image compression and tokenization. By leveraging AI technology, TiTok not only addresses existing challenges but also paves the way for more efficient and effective methods of storing and processing images on the blockchain.