Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
Affiliation:
3
0
4
Pretrained video diffusion models can act as universal, plug-and-play restoration layers for legacy codecs, outperforming VTM-23.0 on both perceptual metrics and classical PSNR while slashing bitrates by up to 94%.
ContextCodec achieves a remarkable balance between quality and intelligibility at ultra-low bitrates, outperforming existing codecs by focusing on content-driven context features.
Get 10x faster generative image compression on GPUs with ProGIC, a lightweight RVQ codec that doesn't sacrifice perceptual quality.