arXiv Computer Vision
Learning Representations to Guide Pixel Diffusion Transformers
Read the original on arXiv Computer Vision →The Flow has not summarised this story yet — read it at arXiv Computer Vision.
Related stories
arXiv Computer Vision
Sep 1
WiT: Waypoint Diffusion Transformers for Alleviating Trajectory Conflict in Pixel-Space Image Generation
arXiv:2603.15132v3 Announce Type: replace Abstract: While recent Flow Matching models avoid the reconstruction bottlenecks of latent autoencoders by operating directly in pixel space, the raw pixel m...
arXiv Machine Learning
4d ago
Residual-Stream Burden Shapes Representation Learning in Diffusion Transformers
The paper investigates why Diffusion Transformers perform well when predicting clean data but poorly when predicting noise or velocity. It introduces the concept of *residual-stream burden*, explaining that noisy targets force the residual stream to carry noise-dependent variation through the network, limiting representation flexibility. Experiments show that spectral concentration of clean targets reduces this burden, and the authors propose Spatially Indexed Hyper-Connections (SiHC) to expand residual bandwidth, achieving state‑of‑the‑art FID on ImageNet.
arXiv Computer Vision
Sep 22
PixelDiT2: Representation-Grounded Pixel Diffusion Transformers
arXiv:2609.24919v1 Announce Type: new Abstract: Recent advances in pixel-space diffusion models have narrowed the image quality gap with latent-space diffusion, but still converge more slowly and lag...
arXiv Computer Vision
Oct 1
DC-SAE: Deep Compression Semantic Autoencoder for Faster Diffusion Convergence
arXiv:2609.39222v1 Announce Type: new Abstract: High-compression tokenizers are essential for scaling latent image generative models. However, aggressive compression creates a fundamental tradeoff be...
arXiv Machine Learning
Aug 4
UDT: Reconciling U-Nets and Diffusion Transformers with Data-Adaptive Token Reduction
arXiv:2608. 01298v1 Announce Type: cross Abstract: Diffusion Transformers (DiTs) have emerged as a core architecture in generative modeling due to their scalability and adaptability to multimodal tasks.
arXiv Computer Vision
4d ago
Pooling Representation Autoencoders for Efficient Diffusion
arXiv:2610.09242v1 Announce Type: new Abstract: Representation Autoencoders (RAEs) generate images from pre-trained visual fea- tures, but their dense token grids make generative modeling expensive....