AI Paper Atlas

Research map / Video / Video segmentation and tracking

SparkVSR: Interactive Video Super-Resolution via Sparse Keyframe Propagation

Jiongze Yu, Xiangbo Gao, Pooja Verlani, Akshay Gadde, Yilin Wang, Balu Adsumilli, Zhengzhong Tu

ECCV 2026 · Video › Video segmentation and tracking · approach: Real-world video super-resolution · 5 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
raise reference-free guidance scale from 0 to 1.5 (Nano-Banana-Pro ref) (ablation)SparkVSR · UDM10PSNR: 26.62 -> 20.89Hurt
raise reference-free guidance scale from 0.5 to 1.5 (Nano-Banana-Pro ref) (ablation)SparkVSR · UDM10MUSIQ: 58.76 -> 68.36Improved
raise reference-free guidance scale from 0.5 to 1.5 (PiSA-SR ref) (ablation)SparkVSR · UDM10MUSIQ: 60.17 -> 70.39Improved
raise reference-free guidance scale from 0.5 to 1.5 (PiSA-SR ref) (ablation)SparkVSR · UDM10PSNR: 25.66 -> 20.94Hurt
raise reference-free guidance scale from 0.5 to 1.5 (Nano-Banana-Pro ref) (ablation)SparkVSR · UDM10DOVER: 0.571 -> 0.746Improved
add pixel-space stage 2 after latent stage 1 (ablation)SparkVSR · UDM10perceptual metricsImproved
add pixel-space stage 2 after latent stage 1 (zero reference) (ablation)SparkVSR · UDM10PSNR: 26.73 -> 26.62Mixed
add a single reference keyframe (ablation)SparkVSR · MovieLQMUSIQ: 56.34 -> 61.73Improved
add more distributed reference frames (three uniform or four I-frames) (ablation)SparkVSR · MovieLQperceptual qualityImproved
use Nano-Banana-Pro reference with the no-reference modelSparkVSR · UDM10DOVER: 0.5494 -> 0.6902Improved
use PiSA-SR reference with the no-reference modelSparkVSR · MovieLQMUSIQ: 56.34 -> 68.88Improved

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