I'm an MSc student studying Computer Science at Tel-Aviv University under the supervision of Prof. Daniel Cohen-Or. I am particularly interested in Computer Vision and am currently working on research focusing on image generation and image manipulation.
StyleFusion: A Generative Model for Disentangling Spatial Segments
We present StyleFusion, a new mapping network for learning a semantically-aware disentangled representation of images. StyleFusion allows users to fuse a set of input latent codes into a single unified image in which each semantic region is controlled by one of the input latent codes, providing plug-and-play flexibiliy in the synthesis process.
ReStyle: A Residual-Based StyleGAN Encoder via Iterative Refinement ICCV, 2021
Recent encoders have made great progress in inverting real images, but they still fall short at times. What if we could give the encoder a second chance? With our method, ReStyle, we try to do exactly that by gradually improving the inversion in a self-correcting manner.
Only a Matter of Style: Age Transformation Using a Style-Based Regression Model SIGGRAPH, 2021
We present SAM, a method for modeling fine-grained life-long age transformation. SAM leverages the rich semantics of the StyleGAN latent space and learns non-linear paths for transforming real face images.
Designing an Encoder for StyleGAN Image Manipulation SIGGRAPH, 2021
Identifying the existence of the distortion-editability and distortion-perception tradeoffs within the StyleGAN latent space, we suggest principles for designing encoders for facilitating editing on real images by balancing these tradeoffs.
Encoding in Style: a StyleGAN Encoder for Image-to-Image Translation CVPR, 2021
We present pixel2style2pixel (pSp), a generic image-to-image translation framework based on an encoder that directly maps real images into the latent space of a pretrained StyleGAN generator.
|September 1, 2021||Metabob Podcast (video can be found here)|
|August 11, 2021||Simon Fraser University Computer Graphics Seminar: "Recent Advancements in GAN Inversion"|
|August 9, 2021||SIGGRAPH 2021 Labs Technical Papers Live Demo on "Only a Matter of Style"|
|June 29, 2021||Adobe Computer Graphics Seminar: "Recent Advancements in GAN Inversion"|
Articles and Other Publications
Introducing the pixel2style2pixel (pSp) Framework with W&B
In this article, we explore the pixel2style2pixel (pSp) framework and show how integrating and tracking your pSp experiments with Weights and Biases is seamless.