With the explosive growth of the Internet of Things (IoT), simultaneous wireless information and power transfer (SWIPT) has emerged as a core solution for powering low-energy devices. To address the limitations of single-user adaptation and the high cost and power consumption caused by high local oscillator (LO) frequencies of up to 60 GHz, this paper proposes a low LO frequency resonant beam system (LLF-RBS) for multi-user self-aligning SWIPT. Our three-mixer phase conjugation (TMPC) circuit architecture reduces the requirement for key LO frequencies to the 28 and 32 GHz bands, representing an approximate 50% reduction. This architecture enables simultaneous phase conjugation, frequency translation, and information modulation within a low-cost hardware envelope. We theoretically prove that the system’s multi-user beamforming mechanism is physically equivalent to a power iteration algorithm, guaranteeing spontaneous convergence to the channel’s principal eigenmodes without explicit channel state information or digital beamforming control. Simulation results demonstrate that the LLF-RBS achieves adaptive self-alignment and watt-level power transfer for multiple users, attaining a downlink spectral efficiency of up to 21.4 bps/Hz.

Due to control limitations in the denoising process and the lack of training, zero-shot video editing methods often struggle to meet user instructions, resulting in generated videos that are visually unappealing and fail to fully satisfy expectations. To address this problem, we propose Align-A-Video, a video editing pipeline that incorporates human feedback through reward fine-tuning. Our approach consists of two key steps: 1) Deterministic Reward Fine-tuning. To reduce optimization costs for expected noise distributions, we propose a deterministic reward tuning strategy. This method improves tuning stability by increasing sample determinism, allowing the tuning process to be completed in minutes; 2) Feature Propagation Across Frames. We optimize a selected anchor frame and propagate its features to the remaining frames, improving both visual quality and semantic fidelity. This approach avoids temporal consistency degradation from reward optimization. Extensive qualitative and quantitative experiments confirm the effectiveness of using reward fine-tuning in Align-A-Video, significantly improving the overall quality of generated videos.