In short, The works hang at the size they actually are.
On an artist website the work arrives as a grid of small squares. A two metre canvas and a study that fits in your hand come out the same size on the screen, and the only thing telling them apart is a line of small type underneath. The Exhibition Hall turns that around. The visitor walks down a room and the works hang at the size they actually are. The large one is across the room before you reach it; the small one is something you have to walk up to.
It opens in a browser. No app, no plugin, and it works on a phone with nothing to install. Every work keeps its title, its medium and its measurements on the wall beside it. Each painter, or each series, has a room of its own, and a visitor can go straight to the one they came for.
The example here holds five painters and forty works: Monet, Cézanne, Klimt, van Gogh and Schiele. Everything in it is in the public domain and the photographs are museum scans published on Wikimedia Commons. No living artist's work was used for a demonstration.
The hall is written with Three.js. During installation every image is converted into three webp derivatives and the visitor receives the one that suits their screen. The work data sits in SQLite and the hall is served from its own address on nginx. The build chain runs end to end: images are fetched from the work list, derivatives are produced, the database is filled and the server entry is written in one pass.
It is built for painters and galleries. Ninety dollars to build, with the first year included, and sixty dollars a year from the second year on. You send the list of works and I build the hall. Before it goes live you walk through it and tell me what should change: the wall colour, the lighting, the order of the works, how many hang on one wall.