cs.CVNov 12, 2026

TokenStream: Multi-Subject Personalized Video Generation with Token-wise LoRA and Guided Attention

Authors: Hung Nguyen, Quang Qui-Vinh Nguyen, Khoi Nguyen

Organizations: National University of Singapore, Singapore · Qualcomm AI Research

Abstract

We introduce TokenStream, an optimization-based framework that composes independently learned subject identities in text-to-video generation without a reference-video training dataset or joint optimization of each new subject combination. Learning from a single reference image makes subject appearance difficult to separate from static-image bias, while merging independently trained adapters can mix identities. Token-wise LoRA (ToRA) restricts low-rank updates to the cross-attention value projections of subject-associated text tokens, providing a separate adaptation path for each subject. A T2V-guided stream copies queries and keys from the frozen base model during early denoising to reuse its layout and motion priors while retaining personalized values. We also introduce Multi-DB, a 30-subject, 122-prompt benchmark with appearance-free prompts and varied interactions. On Multi-DB at 1.3B, TokenStream improves identity preservation from 0.422 for DreamBooth to 0.467 and textual alignment from 0.652 to 0.688.

Figures & tables

Appendix figures & tables9 assets

Supplementary material from the paper’s appendix.

Appendix

Explore similar work

CardsList
  1. LumosX: Relate Any Identities with Their Attributes for Personalized Video Generation

    Mar 20, 2026Jiazheng Xing, Fei Du, Hangjie Yuan +7Video GenerationControllable Video Generation

  2. Custom Forcing: Training-Free Subject Customization for Autoregressive Video Generation

    Oct 2, 2026Yunseung Ok, Hyunsoo Kim, Minseo Kim +1Identity-Preserving Video GenerationStreaming Video Generation