Search papers, labs, and topics across Lattice.
This paper introduces a novel approach called Synthetic Self-Guidance (SSG) for pixel-space diffusion models, where a frozen, pretrained model guides itself using its own generated samples. By leveraging intermediate layer outputs to create coarse predictions and refining these with a lightweight prediction head, the method significantly enhances generation quality while requiring less than 1% of the compute needed for full model training. The results show a remarkable reduction in Fr茅chet Inception Distance (FID) scores, achieving over 50% improvement across various models without the need for real images in training.
A frozen pixel diffusion model can self-guide its generation process using its own samples, achieving over 50% improvement in image quality with minimal additional compute.
Pixel-space diffusion models aim to learn an end-to-end generator directly over raw pixels. This is challenging because a single model must capture both global structure and local texture in the same high-dimensional space. While recent work improves pixel diffusion through alternative prediction targets, training objectives, and architectures, these advances typically require training a new model from scratch. We show there is a cheaper, complementary strategy: \textbf{a frozen, pretrained pixel diffusion model can guide itself}. Our key observation is that intermediate layers of a pretrained pixel diffusion transformer can be decoded into coarse predictions that capture the main low-frequency structure, while the final layers progressively refine local, high-frequency details. We therefore attach a lightweight prediction head to an intermediate layer, keep the backbone frozen, and use the discrepancy between the intermediate and final predictions as a self-guidance direction during sampling. To train this head, we further find that real images are not necessary. Instead, model-generated samples suffice and even outperform real images for training the head, especially in enhancing the high-frequency components that pixel diffusion tends to underfit. Across multiple pixel diffusion models on ImageNet, our \textbf{Synthetic Self-Guidance (SSG)} consistently improves generation while adapter training requires less than 1$\%$ of full-model training compute: it reduces FID by over 50$\%$ across the evaluated JiT variants without classifier-free guidance (CFG) and further improves strong baselines with CFG, e.g., JiT-H/16 from 1.86 to 1.67 and PixelREPA-H/16 from 1.81 to 1.59. Our code is available at https://github.com/zfu006/SSG.
