Research map / Image generation / GANs
StackGAN: Text to Photo-Realistic Image Synthesis With Stacked Generative Adversarial Networks
Han Zhang, Tao Xu, Hongsheng Li, Shaoting Zhang, Xiaogang Wang, Xiaolei Huang, Dimitris N. Metaxas
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 Stage-II refinement GAN on top of Stage-I (256x256) (ablation) | StackGAN · CUB | inception score: 3.02 -> 3.70 | Improved |
| add Conditioning Augmentation to 256x256 StackGAN (ablation) | StackGAN 256x256 · CUB | inception score: 3.31 -> 3.70 | Improved |
| add Conditioning Augmentation to 256x256 Stage-I GAN (ablation) | Stage-I GAN · CUB | inception score: 2.48 -> 3.02 | Improved |
| add Conditioning Augmentation to 64x64 Stage-I GAN (ablation) | Stage-I GAN · CUB | inception score: 2.66 -> 2.95 | Improved |
| add Conditioning Augmentation to 128x128 StackGAN (ablation) | StackGAN 128x128 · CUB | inception score: 3.20 -> 3.35 | Improved |
| input text again at Stage-II (256x256) (ablation) | StackGAN 256x256 · CUB | inception score: 3.45 -> 3.70 | Improved |
| input text again at Stage-II (128x128) (ablation) | StackGAN 128x128 · CUB | inception score: 3.13 -> 3.35 | Improved |
| increase output resolution from 128x128 to 256x256 (ablation) | StackGAN · CUB | inception score: 3.35 -> 3.70 | Improved |
| StackGAN versus GAN-INT-CLS on CUB | StackGAN · CUB | inception score: 2.88 -> 3.70 | Improved |
| StackGAN versus GAN-INT-CLS on Oxford-102 | StackGAN · Oxford-102 | inception score: 2.66 -> 3.20 | Improved |
| StackGAN versus GAWWN on CUB | StackGAN · CUB | inception score: 3.62 -> 3.70 | Improved |
| StackGAN versus GAN-INT-CLS on MS COCO | StackGAN · MS COCO | inception score: 7.88 -> 8.45 | Improved |
Later papers that compared with this method
16 later accepted papers report StackGAN++ as a baseline in their results tables; 14 of them report a better number on most of the results they share with it.
- Text to Image Generation With Semantic-Spatial Aware GAN (CVPR 2022 · better on 0 of 1 shared result)
- MCL-GAN: Generative Adversarial Networks with Multiple Specialized Discriminators (NeurIPS 2022 · better on 1 of 1 shared result)
- CookGAN: Causality Based Text-to-Image Synthesis (CVPR 2020 · better on 3 of 3 shared results)
- Improving Generative Adversarial Networks With Local Coordinate Coding (TPAMI 2020 · better on 2 of 2 shared results)
- Controllable Text-to-Image Generation (NeurIPS 2019 · better on 6 of 6 shared results)
- Dual Adversarial Inference for Text-to-Image Synthesis (ICCV 2019 · better on 6 of 6 shared results)
- Perceptual Pyramid Adversarial Networks for Text-to-Image Synthesis (AAAI 2019 · better on 4 of 4 shared results)
- Learn, Imagine and Create: Text-to-Image Generation from Prior Knowledge (NeurIPS 2019 · better on 3 of 3 shared results)
8 more not listed.
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
- StackGAN++: Realistic Image Synthesis with Stacked Generative Adversarial Networks (TPAMI 2018 · similarity 0.95)
- DF-GAN: A Simple and Effective Baseline for Text-to-Image Synthesis (CVPR 2022 · similarity 0.91)
- Generative Adversarial Text to Image Synthesis (ICML 2016 · similarity 0.89)
- Perceptual Pyramid Adversarial Networks for Text-to-Image Synthesis (AAAI 2019 · similarity 0.89)
- Adversarial Text to Continuous Image Generation (CVPR 2024 · similarity 0.88)
- Exploring Sparse MoE in GANs for Text-conditioned Image Synthesis (CVPR 2025 · 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.
