A transformer model for sparse-view computed tomography reported leading image-quality scores in fixed and variable-view tests and was also tested on one real nanoscale mouse-brain pillar.