Research map / Image generation / Autoregressive and latent generation
PanoLlama: Generating Endless and Coherent Panoramas with Next-Token-Prediction LLMs
Teng Zhou, Xiaoyu Zhang, Yongchuan Tang
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 |
|---|---|---|---|
| use PanoLlama next-crop prediction instead of best baseline MAD | PanoLlama · Panorama 512x5120 | LPIPS: 0.520 -> 0.410 (21.15%) | Improved |
| use PanoLlama instead of MAD | PanoLlama · Panorama 512x5120 | TV: 0.040 -> 0.021 (47.50%) | Improved |
| use PanoLlama instead of MAD | PanoLlama · Panorama 512x5120 | SSIM: 0.268 -> 0.305 (13.81%) | Improved |
| use PanoLlama instead of MAD | PanoLlama · Panorama 512x5120 | DISTS: 0.209 -> 0.196 (6.22%) | Improved |
| use PanoLlama instead of MAD | PanoLlama · Panorama 512x5120 | FID: 40.09 vs 59.23 | Hurt |
| use PanoLlama instead of MultiDiffusion | PanoLlama · Panorama 512x5120 | FID: 39.30 -> 40.09 | Mixed |
| use PanoLlama instead of SMD for speed | PanoLlama · Panorama 512x5120 | time: 241.06s -> 726.33s | Hurt |
| use PanoLlama instead of SyncDiffusion for speed | PanoLlama · Panorama 512x5120 | time: 7232.70s -> 726.33s | Improved |
| use PanoLlama (LlamaGen) with SDXL-based baselines | PanoLlama · Panorama 512x5120 (SDXL baselines) | LPIPS: 0.517 -> 0.410 | Improved |
| increase expansion stride u from 1/8 toward 1 (ablation) | PanoLlama · Panorama stride sweep | COH | No clear effect |
| increase panorama width w' from 1024 to 5120 (ablation) | PanoLlama · Panorama size sweep | COH | No clear effect |
| blend boundary token embeddings with lambda below 1.0 under a single prompt (ablation) | PanoLlama · Panorama single prompt | coherence | Hurt |
Closest papers by meaning
- Conditional Panoramic Image Generation via Masked Autoregressive Modeling (NeurIPS 2025 · similarity 0.87)
- DiffPano: Scalable and Consistent Text to Panorama Generation with Spherical Epipolar-Aware Diffusion (NeurIPS 2024 · similarity 0.87)
- HORIZON: High-Resolution Semantically Controlled Panorama Synthesis (AAAI 2024 · similarity 0.85)
- Taming Stable Diffusion for Text to 360 Panorama Image Generation (CVPR 2024 · similarity 0.85)
- CubeDiff: Repurposing Diffusion-Based Image Models for Panorama Generation (ICLR 2025 · similarity 0.84)
- TreeMeshGPT: Artistic Mesh Generation with Autoregressive Tree Sequencing (CVPR 2025 · similarity 0.84)
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.
