PixelDiT
PulseAugur coverage of PixelDiT — every cluster mentioning PixelDiT across labs, papers, and developer communities, ranked by signal.
2 day(s) with sentiment data
-
New Wiener-Informed Decoding method speeds up AI image model training
Researchers have developed Wiener-informed Decoding (WiD), a novel method to improve the training of pixel-space flow models. By incorporating coarse second-order image statistics, WiD accelerates training and enhances …
-
BudgetPix framework enables adaptive compute for image diffusion models
Researchers have introduced BudgetPix, a novel framework for image diffusion models that adaptively allocates computational resources based on visual complexity. This approach allows a single model checkpoint to operate…
-
ShatterQuant enables block-wise mixed-precision for transformer hardware
Researchers have developed ShatterQuant, a novel framework for mixed-precision quantization in neural networks that allows for independent bit-width assignments to different blocks within a single tensor. This approach …
-
MOSAIK framework boosts image generation efficiency with adaptive patch sizing
Researchers have developed MOSAIK, a novel framework designed to enhance the efficiency of image generation in pixel-space diffusion models. This method addresses the computational cost associated with high token counts…
-
PID method enhances Krea 2 image generation quality
A new method called PID (Pixel Diffusion) has been developed to improve image generation quality in Krea 2, which previously used a suboptimal VAE. PID decodes images directly in pixel space, bypassing the VAE to enhanc…
-
ComfyUI-PiD update adds native model support and FP8 precision
A custom node for ComfyUI, named ComfyUI-PiD, has been updated to support native PixelDiT/PiD model loading and FP8 precision. This update removes reliance on older loading methods and integrates with ComfyUI's native m…
-
New Diffusion Transformers Advance Image Generation and Transmission
Researchers are developing new diffusion transformer models for advanced image generation and transmission. One approach, DDM-SSCC, adapts diffusion language models for lossless pixel-level image transmission, outperfor…