We compare our image model with full-resolution FLUX.2 and token-reduction baselines alongside the Level-of-Token layout. These additional examples let readers inspect fine detail and visual artifacts across a broader set of prompts than the paper can show.