Skip to content

MSR: Multiple Subject Reference for Video Generation

Sep 2026 · arXiv.org · Vol abs/2609.18393 · 0 citations · 8 references
Computer Science

TL;DR

MSR (Multiple Subject Reference), a slot-aware conditioning scheme for LTX-based video generation, is presented and development observations suggest reduced reference confusion relative to an earlier continuous-reference baseline.

Abstract

Conditioning a video generator on multiple images requires preserving appearance while associating each reference with its intended role. We present MSR (Multiple Subject Reference), a slot-aware conditioning scheme for LTX-based video generation. Each reference image is independently encoded as a static clip and represented by a separate latent-token group. A compact Fourier-feature multilayer perceptron adds a numeric slot embedding, while slot-dependent temporal offsets modify the group's rotary coordinates. The reference groups are prepended to noisy target tokens and serve as clean context during target-only flow-matching training. We implement this scheme through low-rank adaptation and release the resulting weights and inference workflows. Qualitative examples demonstrate compositions containing distinct characters and referenced environments in realistic and stylized scenes. Development observations suggest reduced reference confusion relative to an earlier continuous-reference baseline, while similar clothing, complex garments, and viewpoint changes remain challenging. We describe the conditioning mechanism, the retained training configuration, and the observed strengths and limitations of the released system. A supplementary audio-reference experiment adds voice conditioning while keeping the visual parameters frozen.

View source

Similar papers

Sep 2026

ORAV: Benchmarking Audio-Video Generation from Multimodal Contexts

ORAV Bench for Omni Reference Audio-Video Generation is introduced, comprising 380 task instances with 2-10 references, 9 semantic roles, and 30 role compositions, and a reference-aware pairwise protocol that prepares visual and auditory evidence, compares the intended contribution of each reference, and checks the ove...

Jia-Cheng Hua, Xiao-Kun Feng, Jia-Qi Hua et al. · 0 citations
Sep 2026

Complementary Retrieval-Augmented Prompting for Consistent Long-Form Video Generation

Quantitative and qualitative evaluations on multi-shot story generation demonstrate that the Complementary Retrieval-Augmented Prompting method consistently outperforms recent-frame, memory-based, and entity-level retrieval baselines in cross-shot consistency and text-controllability.

Xiang-Han Wei, Xiao-Da Yang, Zhi Wang et al. · 1 citation
Sep 2026

TT-VidT: Decoupling the Temporal Axis for Efficient Motion-Centric Video Pretraining

TT-VidT leads Jester, Something-Something V2, ARID, and Diving48 fine-tuning simultaneously, improving over the strongest non-TT row by 54% ~ 121%, while using 48% fewer encoder FLOPs than DisMo and 55% fewer than VideoMAE or V-JEPA2.

Shih-Ying Yeh, Daniel Z. Kaplan, Xue-Hai Wang et al. · 0 citations
#artificial intelligence Preprint Aug 2026

VIBE: Video Instruction-aligned Background music gEneration

VIBE is introduced, a novel text-and-video-to-music (T+V2M) generation model that leverages a depth-wise cross-layer conditioning mechanism that dynamically bridges the planning and diffusion refinement heads and a comprehensive reward modeling taxonomy, optimizing for both hard, verifiable constraints and soft, subjec...

Aryan Vijay Bhosale, Vaibhavi Lokegaonkar, Vishnu Raj et al. · 0 citations
#generative ai Review Sep 2026

Reference Fidelity in AI-Generated Short-Form Video Across Four Prompt Packages

Generative video systems can produce short clips from textual and visual instructions, yet their ability to preserve the content of a human reference remains uncertain. This exploratory study examines 160 videos generated with Doubao-Seedance-2.0 from 40 human-made short-form videos drawn from Douyin and Xiaohongshu. T...

Chloe Zihan Jin · 0 citations
Sep 2026

RefCompose: Multi-Reference Image Generation via LoRA-Conditioned Diffusion

On the Dense Layout protocol, RefCompose consistently outperforms layout based and state of the art multi reference baselines on color, texture, shape, spatial accuracy, and identity/content preservation at higher reference counts, all with constant inference memory, making it a practical building block for multi subje...

Sai Sri Teja Kuppa, Parth Shinde, S. PriyadharsanBalaji et al. · 0 citations

We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.