Diffusion models have demonstrated strong generative performance; however, generated samples often fail to fully align with human intent. This paper studies a test-time scaling method that enables sampling from regions with higher human-aligned reward values. Existing gradient guidance methods approximate the expected future reward (EFR) at an intermediate particle xt using a Taylor approximation, but this approximation at each time step incurs high computational cost due to sequential neural backpropagation. We show that the EFR at any xt can be computed using only marginal samples from a pre-trained diffusion model. The proposed EFR formulation detaches the neural dependency between xt and the EFR, enabling closed-form guidance computation without neural backpropagation. To further improve efficiency, we introduce lookahead sampling to collect marginal samples. For final sample generation, we use an accurate solver that guides particles toward high-reward lookahead samples. We refer to this sampling scheme as LiDAR sampling. LiDAR achieves substantial performance improvements using only three samples with a 3-step lookahead solver, exhibiting steep performance gains as lookahead accuracy and sample count increase; notably, it reaches the same GenEval performance as the latest gradient guidance method for SDXL with a 9.5x speedup.
@article{arxiv.2602.03211,
title = {Lookahead Sample Reward Guidance for Test-Time Scaling of Diffusion Models},
author = {Yeongmin Kim and Donghyeok Shin and Byeonghu Na and Minsang Park and Richard Lee Kim and Il-Chul Moon},
journal= {arXiv preprint arXiv:2602.03211},
year = {2026}
}