NVIDIA researchers have introduced PiD, a novel Pixel Diffusion Decoder that unifies image decoding and upsampling into a single generative module. This model operates directly in high-resolution pixel space, producing super-resolved images in a single pass. Two variants, 2k and 2kto4k, are available, supporting upscaling to 2048px and 4096px respectively, with checkpoints released under a research-only license. AI
IMPACT Enables faster and higher-resolution image generation by unifying decoding and upsampling.
RANK_REASON Model release with associated paper and code. [lever_c_demoted from research: ic=1 ai=1.0]
Read on Hugging Face Trending Models →
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →