Abstract: Despite the unprecedented success of text-to-image diffusion models, controlling the number of depicted objects using text is surprisingly hard. This is important for various applications ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results