Ing. Adéla Šubrtová

We propose a method for detecting and automatically correcting perceptual artifacts on synthetic face images. Recent generative models, such as diffusion models, can produce photorealistic images. However, these models often generate visual defects on the faces of people, especially at low resolutions, which impairs the quality of the images. We use a face detector and a binary classifier to identify perceptual artifacts. The classifier was trained on our dataset of manually annotated synthetic face images generated by a diffusion model, half of which contain perceptual artifacts. We compare our method with several baselines and show that it achieves superior accuracy of 93% on an independent test set. In addition, we propose a simple mechanism for automatically correcting the distorted faces using inpainting. For each face with artifact response, we generate several replacement candidates by inpainting and choose the best one by the lowest artifact score. The best candidate is then back-projected into to the image. Inpainting ensures a seamless connection between the corrected face and the original image. Our method improves the realism and quality of synthetic images.

In this paper we present Diffusion Image Analogies—an examplebased image editing approach that builds upon the concept of image analogies originally introduced by Hertzmann et al. [2001]. Given a pair of images that specify the intent of a specific transition, our approach enables to modify the target image in a way that it follows the analogy specified by this exemplar. In contrast to previous techniques which were able to capture analogies mostly on the low-level textural details our approach handles also changes in higher level semantics including transition of object domain, change of facial expression, or stylization. Although similar modifications can be achieved using diffusion models guided by text prompts [Rombach et al. 2022] our approach can operate solely in the domain of images without the need to specify the user’s intent using textual form.We demonstrate power of our approach in various challenging scenarios where the specified analogy would be difficult to transfer using previous techniques.

We present ChunkyGAN—a novel paradigm for modeling and editing images using generative adversarial networks. Unlike previous techniques seeking a global latent representation of the input image, our approach subdivides the input image into a set of smaller components (chunks) specified either manually or automatically using a pre-trained segmentation network. For each chunk, the latent code of a generative network is estimated locally with greater accuracy thanks to a smaller number of constraints. Moreover, during the optimization of latent codes, segmentation can further be refined to improve matching quality. This process enables high-quality projection of the original image with spatial disentanglement that previous methods would find challenging to achieve. To demonstrate the advantage of our approach, we evaluated it quantitatively and also qualitatively in various image editing scenarios that benefit from the higher reconstruction quality and local nature of the approach. Our method is flexible enough to manipulate even out-of-domain images that would be hard to reconstruct using global techniques.

We propose a neural network which takes two inputs, a hair image and a face image, and produces an output image having the hair of the hair image seamlessly merged with the inner face of the face image. Our architecture consists of neural networks mapping the input images into a latent code of a pretrained StyleGAN2 which generates the output high-definition image. We propose an algorithm for training parameters of the architecture solely from synthetic images generated by the StyleGAN2 itself without the need of any annotations or external dataset of hairstyle images. We empirically demonstrate the effectiveness of our method in applications including hair-style transfer, hair generation for 3D morphable models, and hair-style interpolation. Fidelity of the generated images is verified by a user study and by a novel hairstyle metric proposed in the paper.