Abstract: This paper presented an approach for evaluating image generation models by measuring their consistency in producing similar results across multiple prompts and seeds. We proposed a method ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results