StyleMC: Multi-Channel Based Fast Text-Guided Image Generation and Manipulation

StyleGAN-XL: Scaling StyleGAN to Large Diverse Datasets

“…However, techniques using the StyleSpace [61], e.g. StyleMC [31], tend to yield better results in our experiments confirming the findings of concurrent work by Alaluf et al [4]. Furthermore, we remark that for applications where equivariance is not essential, our framework can easily be used with StyleGAN2 layers instead.…”

“…6 (Left), we first invert a given source image via latent space optimization. We then apply different manipulation directions obtained by GANspace [20] and StyleMC [31]. Prior work [23] also investigates in-plane translation.…”

Text and Image Guided 3D Avatar Generation and Manipulation

“…A more recent work, LatentCLR [60], proposes a contrastive learning approach to find unsupervised directions that are transferable to different classes. Moreover, StyleCLIP [40] and StyleMC [34] both propose using CLIP for text-guided manipulation of both randomly generated and encoded images with StyleGAN2. These methods show that it is possible to use CLIP for fine-grained and disentangled manipulations of images.…”

Discovering Multiple and Diverse Directions for Cognitive Image Properties

“…Similarly, [1] uses conditional continuous normalization flows to perform supervised attribute processing in the latent space of StyleGAN2. Recently, text-based manipulation methods have been proposed [15,21], which use CLIP [24] to perform fine-grained and disentangled manipulations of images.…”

“…However, most previous work on the manipulation of latent space has focused on the discovery of a single direction by which the desired editing can be achieved, such as zoom-in [6,11] or altering facial features, such as applying a lipstick [15,34]. In this work, we focus on the discovery of multiple and diverse directions that can achieve the desired edit.…”