# JVT: A Simplified DiT Transformer Tackles Pixel-Space Video Diffusion Without Autoencoders

_Research · published 2026-11-19_

Researchers have introduced JVT (Just Video Transformer), a DiT-style Transformer designed for pixel-space video diffusion that removes the dependency on pretrained variational autoencoders, cascaded stages, U-Net skip connections, or auxiliary detailers. By using large spatiotemporal patches under a rectified-flow objective, JVT makes direct modeling of raw video tractable, achieving generation quality competitive with latent diffusion baselines in both class-conditioned and unconditional settings.

A key finding from the work is that patch embedding represents a significant bottleneck for generation quality and efficiency, while standard adaptive layer normalization (AdaLN) remains an effective conditioning design. The pixel-space formulation also unlocks native joint generation of pixel-aligned modalities — the paper demonstrates camera-conditioned RGB-disparity generation on RealEstate10K, improving camera-following accuracy and yielding explicit scene geometry at inference without an additional reconstruction model.

## Sources
- [zhoutimemachine.github.io](https://zhoutimemachine.github.io/JVT/)
---
Canonical: https://genbuzz.news/posts/jvt-a-simplified-dit-transformer-tackles-pixel-space-video-diffusion-without-autoencoders
