Peer-Reviewed | Wenjie Shuhttps://wenjieshu.github.io/tags/peer-reviewed/Peer-ReviewedHugo Blox Builder (https://hugoblox.com)en-usThu, 01 Jan 2026 00:00:00 +0000https://wenjieshu.github.io/media/icon_hu_187844eed2f8f505.pngPeer-Reviewedhttps://wenjieshu.github.io/tags/peer-reviewed/AlignVid: Training-Free Attention Scaling for Semantic Fidelity in Text-Guided Image-to-Video Generationhttps://wenjieshu.github.io/publications/alignvid/Thu, 01 Jan 2026 00:00:00 +0000https://wenjieshu.github.io/publications/alignvid/<p>Accepted at ICML 2026.</p>LoopViT: Scaling Visual ARC with Looped Transformershttps://wenjieshu.github.io/publications/loopvit/Thu, 01 Jan 2026 00:00:00 +0000https://wenjieshu.github.io/publications/loopvit/<p>Accepted at EMNLP 2026 Findings.</p> <!-- Optional inline PDF preview (if pipeline.pdf exists in this folder) --> <div style="margin-top: 1rem;"> <object data="pipeline.pdf" type="application/pdf" width="100%" height="640px"> <p> Can't display the PDF inline. Click to view/download: <a href="pipeline.pdf">pipeline.pdf</a> </p> </object> <p style="text-align: center; color: var(--color-contrast-600); margin-top: 0.5rem;"> Figure: LoopViT overview. </p> </div>ScalingAR: Scaling Confidence for Autoregressive Image Generationhttps://wenjieshu.github.io/publications/go-with-your-gut/Thu, 01 Jan 2026 00:00:00 +0000https://wenjieshu.github.io/publications/go-with-your-gut/<p>Accepted at ICML 2026.</p> <!-- Inline preview for Go with Your Gut figure (PDF). --> <div style="margin-top: 1rem;"> <object data="Figure2.pdf" type="application/pdf" width="100%" height="640px"> <p> Can't display the PDF inline. Click to view/download: <a href="Figure2.pdf">Figure2.pdf</a> </p> </object> <p style="text-align: center; color: var(--color-contrast-600); margin-top: 0.5rem;"> Figure: Confidence scaling overview. </p> </div>Show, Don't Tell: Morphing Latent Reasoning into Image Generationhttps://wenjieshu.github.io/publications/latentmorph/Thu, 01 Jan 2026 00:00:00 +0000https://wenjieshu.github.io/publications/latentmorph/<p>Accepted at ICML 2026.</p> <!-- Inline preview for LatentMorph figure (PDF). --> <div style="margin-top: 1rem;"> <object data="figure.pdf" type="application/pdf" width="100%" height="640px"> <p> Can't display the PDF inline. Click to view/download: <a href="figure.pdf">figure.pdf</a> </p> </object> <p style="text-align: center; color: var(--color-contrast-600); margin-top: 0.5rem;"> Figure: LatentMorph overview. </p> </div>ThinkVid: Benchmarking Visual Reasoning in Video Generative Modelshttps://wenjieshu.github.io/publications/thinkvid/Thu, 01 Jan 2026 00:00:00 +0000https://wenjieshu.github.io/publications/thinkvid/<p>Accepted at CVPR 2026.</p>CMT: Cross Modulation Transformer with Hybrid Loss for Pansharpeninghttps://wenjieshu.github.io/publications/journal-article/Mon, 01 Jul 2024 00:00:00 +0000https://wenjieshu.github.io/publications/journal-article/<p>This work introduces CMT and a hybrid frequency-domain loss to advance pansharpening quality.</p> <!-- Inline preview for CMT figure (PDF). --> <div style="margin-top: 1rem;"> <object data="modulation.pdf" type="application/pdf" width="100%" height="640px"> <p> Can't display the PDF inline. Click to view/download: <a href="modulation.pdf">modulation.pdf</a> </p> </object> <p style="text-align: center; color: var(--color-contrast-600); margin-top: 0.5rem;"> Figure: CMT modulation overview. </p> </div>Exploring the Low-Pass Filtering Behavior in Image Super-Resolutionhttps://wenjieshu.github.io/publications/conference-paper/Wed, 01 May 2024 00:00:00 +0000https://wenjieshu.github.io/publications/conference-paper/<p>This work studies why SR models exhibit low-pass tendencies and discusses remedies.</p>