# Unsupervised Image-to-Image Translation via Pre-trained StyleGAN2 Network

arXiv: https://arxiv.org/abs/2010.05713 · HTML: https://mlsanity.com/paper/2010.05713/

## Key facts

- Cited by 36 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 12 Oct 2020.

## Built on by (most cited first, 30 of 36)

- [CIPS-3D: A 3D-Aware Generator of GANs Based on Conditionally-Independent Pixel Synthesis](https://mlsanity.com/paper/2110.09788/) (2021)
- [Semantic Image Synthesis via Diffusion Models](https://mlsanity.com/paper/2207.00050/) (2022)
- [NeRF-Art: Text-Driven Neural Radiance Fields Stylization](https://mlsanity.com/paper/2212.08070/) (2022)
- [Pastiche Master: Exemplar-Based High-Resolution Portrait Style Transfer](https://mlsanity.com/paper/2203.13248/) (2022)
- [StyleAlign: Analysis and Applications of Aligned StyleGAN Models](https://mlsanity.com/paper/2110.11323/) (2021)
- [BlendGAN: Implicitly GAN Blending for Arbitrary Stylized Face Generation](https://mlsanity.com/paper/2110.11728/) (2021)
- [State-of-the-Art in the Architecture, Methods and Applications of StyleGAN](https://mlsanity.com/paper/2202.14020/) (2022)
- [TryOnGAN: Body-Aware Try-On via Layered Interpolation](https://mlsanity.com/paper/2101.02285/) (2021)
- [Temporally Consistent Semantic Video Editing](https://mlsanity.com/paper/2206.10590/) (2022)
- [Cross-Domain and Disentangled Face Manipulation with 3D Guidance](https://mlsanity.com/paper/2104.11228/) (2021)
- [Unsupervised Image-to-Image Translation with Generative Prior](https://mlsanity.com/paper/2204.03641/) (2022)
- [DynaGAN: Dynamic Few-shot Adaptation of GANs to Multiple Domains](https://mlsanity.com/paper/2211.14554/) (2022)
- [StyleGAN of All Trades: Image Manipulation with Only Pretrained StyleGAN](https://mlsanity.com/paper/2111.01619/) (2021)
- [Fine-Tuning StyleGAN2 For Cartoon Face Generation](https://mlsanity.com/paper/2106.12445/) (2021)
- [StyleDomain: Efficient and Lightweight Parameterizations of StyleGAN for One-shot and Few-shot Domain Adaptation](https://mlsanity.com/paper/2212.10229/) (2022)
- [GP-UNIT: Generative Prior for Versatile Unsupervised Image-to-Image Translation](https://mlsanity.com/paper/2306.04636/) (2023)
- [Towards Robust Event-based Networks for Nighttime via Unpaired Day-to-Night Event Translation](https://mlsanity.com/paper/2407.10703/) (2024)
- [CLIP2GAN: Towards Bridging Text with the Latent Space of GANs](https://mlsanity.com/paper/2211.15045/) (2022)
- [HyperStyle3D: Text-Guided 3D Portrait Stylization via Hypernetworks](https://mlsanity.com/paper/2304.09463/) (2023)
- [AgileGAN3D: Few-Shot 3D Portrait Stylization by Augmented Transfer Learning](https://mlsanity.com/paper/2303.14297/) (2023)
- [Asymmetric Modality Translation For Face Presentation Attack Detection](https://mlsanity.com/paper/2110.09108/) (2021)
- [Fix the Noise: Disentangling Source Feature for Controllable Domain Translation](https://mlsanity.com/paper/2303.11545/) (2023)
- [SwiftAvatar: Efficient Auto-Creation of Parameterized Stylized Character on Arbitrary Avatar Engines](https://mlsanity.com/paper/2301.08153/) (2023)
- [3D Face Style Transfer with a Hybrid Solution of NeRF and Mesh Rasterization](https://mlsanity.com/paper/2311.13168/) (2023)
- [ToonAging: Face Re-Aging upon Artistic Portrait Style Transfer](https://mlsanity.com/paper/2402.02733/) (2024)
- [Pix2Next: Leveraging Vision Foundation Models for RGB to NIR Image Translation](https://mlsanity.com/paper/2409.16706/) (2024)
- [Revealing Directions for Text-guided 3D Face Editing](https://mlsanity.com/paper/2410.04965/) (2024)
- [DIVE: Inverting Conditional Diffusion Models for Discriminative Tasks](https://mlsanity.com/paper/2504.17253/) (2025)
- [ChildPredictor: A Child Face Prediction Framework with Disentangled Learning](https://mlsanity.com/paper/2204.09962/) (2022)
- [Fix the Noise: Disentangling Source Feature for Transfer Learning of StyleGAN](https://mlsanity.com/paper/2204.14079/) (2022)

## Abstract (arXiv)

Image-to-Image (I2I) translation is a heated topic in academia, and it also has been applied in real-world industry for tasks like image synthesis, super-resolution, and colorization. However, traditional I2I translation methods train data in two or more domains together. This requires lots of computation resources. Moreover, the results are of lower quality, and they contain many more artifacts. The training process could be unstable when the data in different domains are not balanced, and modal collapse is more likely to happen. We proposed a new I2I translation method that generates a new model in the target domain via a series of model transformations on a pre-trained StyleGAN2 model in the source domain. After that, we proposed an inversion method to achieve the conversion between an image and its latent vector. By feeding the latent vector into the generated model, we can perform I2I translation between the source domain and target domain. Both qualitative and quantitative evaluations were conducted to prove that the proposed method can achieve outstanding performance in terms of image quality, diversity and semantic similarity to the input and reference images compared to state-of-the-art works.