Our method leverages the comprehensive priors of appearance and geometry offered by image-text diffusion models to generate multi-view images of avatars in various styles. During data generation, we ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results