A portrait prompt shown with its result — copy it, swap the subject, run your own.
Prompt
16:9 select * from artifact_diorama_archives where subject_media = '{argument name="media title" default="[media_title]"}' and base_structure = 'giant_hollowed_out_sphere_cracked_open' and exterior_shell = ( select manuscript_aesthetic, typography from cultural_heritage_db where lore = '{argument name="media title" default="[media_title]"}' format as 'flat_2d_painted_fresco_and_calligraphy_on_outer_curve' ) and interior_payload_strata = array[ (level: 'upper_apex', asset: 'infer_climactic_fortress_or_throne({argument name="media title" default="[media_title]"})'), (level: 'middle_caverns', asset: 'infer_mythological_creatures_and_terrain({argument name="media title" default="[media_title]"})'), (level: 'lower_basin', asset: 'infer_epic_armies_and_waterways({argument name="media title" default="[media_title]"})') ] and materials = 'carved_stone_wood_and_antiqued_gold' and optical_engine = 'macro_photography_depth_of_field';
It leans on a clear description of the subject and mood instead of camera jargon. That makes it a flexible base: add a focal length, a light direction and an aspect ratio and it will land the same way on every run.