AI Paper Atlas

Research map / Video / Video generation

TinyHistory: Lightweight Video History Embeddings via Two-Stage Context Learning

lvmin zhang, Shengqu Cai, Muyang Li, Chong Zeng, Beijia Lu, Anyi Rao, Song Han, Gordon Wetzstein, Maneesh Agrawala

ECCV 2026 · Video › Video generation · approach: Compositional text-to-video diffusion · 33 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add Stage I randomized frame-query pretraining (ablation)TinyHistory 4x4x2 · Storyboard generationChar / FaceID: Char 52.64 -> 80.19; FaceID 58.38 -> 70.71Improved
train the encoder jointly in Stage II instead of freezing it (ablation)TinyHistory 4x4x2 · Storyboard generationChar: 61.42 -> 80.19Improved
add the LR branch to the HR-only encoder (ablation)TinyHistory 4x4x2 · Storyboard generationChar: 67.73 -> 80.19Improved
add the HR branch to the LR-only encoder (ablation)TinyHistory 4x4x2 · Storyboard generationChar: 74.29 -> 80.19Improved
lower spatial/temporal compression (4x4x2 -> 2x2x2) (ablation)TinyHistory · Storyboard generationScene / ELO: Scene 72.21 -> 87.11; ELO 1262 -> 1332Improved
halve context cost from 2x2x1 to 2x2x2 (ablation)TinyHistory · Storyboard generationELO: 1342 -> 1332No clear effect
use 4x temporal instead of 4x spatial compression (2x2x4 vs 4x4x2) (ablation)TinyHistory · Storyboard generationChar: 80.19 -> 90.24Improved
add 3-frame sliding window overlap (ablation)TinyHistory 4x4x2 · Storyboard generationScene: 72.21 -> 82.51Improved
add cross-attention from encoder to DiT (ablation)TinyHistory 4x4x2 · Storyboard generationChar / Obj: Char 80.19 -> 87.01; Obj 74.38 -> 80.75Improved
add a second encoder (2x2x8) (ablation)TinyHistory 4x4x2 · Storyboard generationChar / Obj: Char 80.19 -> 88.25; Obj 74.38 -> 83.33Improved
cache history KV across diffusion steps (ablation)TinyHistory 4x4x2 · Storyboard generationFaceID / Scene: FaceID 70.71 -> 63.61; Scene 72.21 -> 65.28Hurt
weaker history conditioning (no pretraining or frozen encoder) (ablation)TinyHistory · Storyboard generationDynamic degree: 84.31 -> 95.69 (no Stage I)Mixed

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy