Research map / Image generation / Autoregressive and latent generation
SynVAR: Synergizing Spatial and Semantic Alignment in Visual Autoregressive Model
Zhennan Chen, Tianxing Shi, Pengcheng Xu, Kepan Nan, Qian Wang, Zili Yi, Jian Yang, Ying Tai
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 changed | Where | Measured | Effect |
|---|---|---|---|
| add SynVAR to Infinity | Infinity · GenEval | Overall: 55.48 -> 70.79 | Improved |
| add SynVAR to Switti | Switti · GenEval | Overall: 41.09 -> 45.27 | Improved |
| add DenseDiffusion or RAG-Diffusion enhancement | Infinity · GenEval | Overall: 55.48 -> 42.23 / 46.44 | Hurt |
| add global guidance (ablation) | SynVAR · GenEval | Position: 23.75 -> 63.00 | Improved |
| add receptive field constraint (ablation) | SynVAR · GenEval | Counting: 21.25 -> 58.25 | Improved |
| add high-frequency compensation (ablation) | SynVAR · GenEval | Overall: 66.11 -> 70.79 | Improved |
| Gaussian instead of constant or topK mask (ablation) | receptive field constraint · GenEval | Overall: 68.48 / 68.59 -> 70.79 | Improved |
| apply guidance at step 2 only vs steps 1-5 (ablation) | global guidance + receptive field · GenEval | Overall: 64.49 -> 70.79 | Improved |
| raise enhancing ratio delta from 0.01 to 0.03 (ablation) | high-frequency compensation · GenEval | Overall: 70.79 -> 68.04 | Hurt |
| combine SynVAR with FastVAR | FastVAR · GenEval | Overall: 56.99 -> 69.16 | Improved |
| add SynVAR | Switti · GenEval | Position: 14.25 -> 16.25 | No clear effect |
| add high-frequency compensation | VAR inference · inference time | seconds/image: 2.60 -> 3.08 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- Infinity — Infinity: Scaling Bitwise AutoRegressive Modeling for High-Resolution Image Synthesis (CVPR 2025) · better on 3 of 3 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- SSG: Scaled Spatial Guidance for Multi-Scale Visual Autoregressive Generation (ICLR 2026 · similarity 0.90)
- Visual Autoregressive Modeling: Scalable Image Generation via Next-Scale Prediction (NeurIPS 2024 · similarity 0.90)
- ARVAR: Accelerating Visual Autoregressive Model via Attention Retrospect (ECCV 2026 · similarity 0.89)
- FlexVAR: Flexible Visual Autoregressive Modeling without Residual Prediction (NeurIPS 2025 · similarity 0.88)
- Data-efficient Large Vision Models through Sequential Autoregression (ICML 2024 · similarity 0.87)
- Edit the Bits, Diff the Codes: Bitwise Residual Editing for Visual Autoregressive Models (NeurIPS 2026 · similarity 0.87)
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.
