One-step Latent-free Image Generation with Pixel Mean Flows

Yiyang Lu, Susie Lu, Qiao Sun, Hanhong Zhao, Zhicheng Jiang, Xianbang Wang, Tianhong Li, Zhengyang Geng, Kaiming He
Proceedings of the 43rd International Conference on Machine Learning, PMLR 306:81265-81277, 2026.

Abstract

Modern diffusion/flow-based models for image generation typically exhibit two core characteristics: (i) using multi-step sampling, and (ii) operating in a latent space. Recent advances have made encouraging progress on each aspect individually, paving the way toward one-step diffusion/flow without latents. In this work, we take a further step towards this goal and propose "pixel MeanFlow" (pMF). Our core guideline is to formulate the network output space and the loss space separately. The network target is designed to be on a presumed low-dimensional image manifold (i.e., x-prediction), while the loss is defined via MeanFlow in the velocity space. We introduce a simple transformation between the image manifold and the average velocity field. In experiments, pMF achieves strong results for one-step latent-free generation on ImageNet at 256$\times$256 resolution (2.22 FID) and 512$\times$512 resolution (2.48 FID), filling a key missing piece in this regime. We hope that our study will further advance the boundaries of diffusion/flow-based generative models.

Cite this Paper

BibTeX
@InProceedings{pmlr-v306-lu26t, title = {One-step Latent-free Image Generation with Pixel Mean Flows}, author = {Lu, Yiyang and Lu, Susie and Sun, Qiao and Zhao, Hanhong and Jiang, Zhicheng and Wang, Xianbang and Li, Tianhong and Geng, Zhengyang and He, Kaiming}, booktitle = {Proceedings of the 43rd International Conference on Machine Learning}, pages = {81265--81277}, year = {2026}, editor = {Zhang, Tong and Dudik, Miroslav and Jaggi, Martin and Agarwal, Alekh and Li, Sharon and Schuurmans, Dale and Zhu, Jerry and Berkenkamp, Felix and Dong, Hanze and Bietti, Alberto}, volume = {306}, series = {Proceedings of Machine Learning Research}, month = {06--11 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v306/main/assets/lu26t/lu26t.pdf}, url = {https://proceedings.mlr.press/v306/lu26t.html}, abstract = {Modern diffusion/flow-based models for image generation typically exhibit two core characteristics: (i) using multi-step sampling, and (ii) operating in a latent space. Recent advances have made encouraging progress on each aspect individually, paving the way toward one-step diffusion/flow without latents. In this work, we take a further step towards this goal and propose "pixel MeanFlow" (pMF). Our core guideline is to formulate the network output space and the loss space separately. The network target is designed to be on a presumed low-dimensional image manifold (i.e., x-prediction), while the loss is defined via MeanFlow in the velocity space. We introduce a simple transformation between the image manifold and the average velocity field. In experiments, pMF achieves strong results for one-step latent-free generation on ImageNet at 256$\times$256 resolution (2.22 FID) and 512$\times$512 resolution (2.48 FID), filling a key missing piece in this regime. We hope that our study will further advance the boundaries of diffusion/flow-based generative models.} }
Endnote
%0 Conference Paper %T One-step Latent-free Image Generation with Pixel Mean Flows %A Yiyang Lu %A Susie Lu %A Qiao Sun %A Hanhong Zhao %A Zhicheng Jiang %A Xianbang Wang %A Tianhong Li %A Zhengyang Geng %A Kaiming He %B Proceedings of the 43rd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2026 %E Tong Zhang %E Miroslav Dudik %E Martin Jaggi %E Alekh Agarwal %E Sharon Li %E Dale Schuurmans %E Jerry Zhu %E Felix Berkenkamp %E Hanze Dong %E Alberto Bietti %F pmlr-v306-lu26t %I PMLR %P 81265--81277 %U https://proceedings.mlr.press/v306/lu26t.html %V 306 %X Modern diffusion/flow-based models for image generation typically exhibit two core characteristics: (i) using multi-step sampling, and (ii) operating in a latent space. Recent advances have made encouraging progress on each aspect individually, paving the way toward one-step diffusion/flow without latents. In this work, we take a further step towards this goal and propose "pixel MeanFlow" (pMF). Our core guideline is to formulate the network output space and the loss space separately. The network target is designed to be on a presumed low-dimensional image manifold (i.e., x-prediction), while the loss is defined via MeanFlow in the velocity space. We introduce a simple transformation between the image manifold and the average velocity field. In experiments, pMF achieves strong results for one-step latent-free generation on ImageNet at 256$\times$256 resolution (2.22 FID) and 512$\times$512 resolution (2.48 FID), filling a key missing piece in this regime. We hope that our study will further advance the boundaries of diffusion/flow-based generative models.
APA
Lu, Y., Lu, S., Sun, Q., Zhao, H., Jiang, Z., Wang, X., Li, T., Geng, Z. & He, K.. (2026). One-step Latent-free Image Generation with Pixel Mean Flows. Proceedings of the 43rd International Conference on Machine Learning, in Proceedings of Machine Learning Research 306:81265-81277 Available from https://proceedings.mlr.press/v306/lu26t.html.

Related Material