Research map / Image generation / GANs
Spiral Generative Network for Image Extrapolation
Dongsheng Guo, Hongzhi Liu, Haoru Zhao, Yunhao Cheng, Qingwei Song, Zhaorui Gu, Haiyong Zheng, Bing Zheng
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 |
|---|---|---|---|
| generate slices in spiral turns rather than one-by-one directions (ablation) | SpiralNet · Stanford Cars (four-side) | PSNR: 14.04 -> 14.31 | Improved |
| add closest slice input to the slice generator (ablation) | SpiralNet · Stanford Cars (four-side) | PSNR: 14.02 -> 14.31 | Improved |
| use closest slice instead of sub-image slice as adjacent-turn dependency (ablation) | SpiralNet · Stanford Cars (four-side) | PSNR: 14.08 -> 14.31 | Improved |
| generate slices of one turn sequentially instead of simultaneously (ablation) | SpiralNet · Stanford Cars (four-side) | PSNR: 14.18 -> 14.31; FID 21.96 -> 23.64 | Mixed |
| clockwise instead of anticlockwise slice generation (ablation) | SpiralNet · Stanford Cars (four-side) | PSNR: 14.27 -> 14.31 | No clear effect |
| add sub-image input via AdaIN to the baseline SliceGAN (ablation) | SliceGAN · Stanford Cars (four-side) | PSNR: 13.95 -> 14.02 | Improved |
| add closest slice input via SPADE to the baseline SliceGAN (ablation) | SliceGAN · Stanford Cars (four-side) | PSNR: 13.95 -> 14.18 | Improved |
| use ternary inputs in the proper roles rather than exchanging imaginary and closest slices (ablation) | SliceGAN · Stanford Cars (four-side) | PSNR: 13.98 -> 14.31 | Improved |
| use slice size 32 instead of 4 (ablation) | SliceGAN · Stanford Cars (four-side) | PSNR: 13.58 -> 14.31 | Improved |
| use slice size 32 instead of 64 (ablation) | SliceGAN · Stanford Cars (four-side) | PSNR: 14.05 -> 14.31 | Improved |
| add hue-color loss to ImagineGAN versus no hue-color loss (ablation) | ImagineGAN · Flowers | PSNR: 10.49 -> 14.98 | Improved |
| hue-color loss instead of color loss in ImagineGAN (ablation) | ImagineGAN · Stanford Cars | PSNR: 10.75 -> 13.56 | Improved |
Closest papers by meaning
- Plug & Play Generative Networks: Conditional Iterative Generation of Images in Latent Space (CVPR 2017 · similarity 0.86)
- SinGAN: Learning a Generative Model From a Single Natural Image (ICCV 2019 · similarity 0.85)
- PetsGAN: Rethinking Priors for Single Image Generation (AAAI 2022 · similarity 0.85)
- Generative Photography: Scene-Consistent Camera Control for Realistic Text-to-Image Synthesis (CVPR 2025 · similarity 0.83)
- ExtraNeRF: Visibility-Aware View Extrapolation of Neural Radiance Fields with Diffusion Models (CVPR 2024 · similarity 0.83)
- Learning to Manipulate Artistic Images (AAAI 2024 · similarity 0.83)
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.
