The curator
Give it a collection and it looks at every work — subject, palette, handling, what the thing seems to be about — and then proposes exhibitions built out of what the works have in common. Give it a theme and it works to the theme. Give it nothing and it has to find its own reason, which is the harder and more interesting half.
It reads the chain rather than a marketplace, so any ERC-721 works — a Manifold contract, a Transient Labs contract, your own — and Tezos comes through Objkt. It also reads the Met, Cleveland and the Art Institute, who give their collections away. Nothing is minted, nothing is bought, and nothing you point it at is stored beyond a description of the pictures.
Or give it a person instead of an address. Through Raster, an artist is one handle and you get their whole body of work across every chain they publish on — XCOPY is four wallets on two chains, and this is one box. A collection is one wallet you already know, which Raster turns into every other wallet on that profile: the vault and the hot wallet, without anybody having to paste both.
Click any work and it finds that work’s kin, answering twice: the closest things by other artists working now, and the closest things in a museum. The second is the more interesting one — a glitched self-portrait minted last week has ancestors, and they are four hundred years old and free to look at. The pool grows as the tool is used: every work it reads becomes something a later work can be paired with.
Or start from an image you have. Upload a photograph of a painting, a screenshot, something unfinished — it gets looked at the same way, and if nothing already read resembles it the tool goes and searches the museums on what it saw, reads what comes back, and answers from that. Nothing needs to be minted, or on a chain, or done.
An uploaded image is not pinned, not stored, and never joins the pool anyone else searches; it is resized in your browser first, so the camera, the timestamp and the location never leave your machine. And only published work joins that pool at all — a contract or an artist’s own catalogue is published, a wallet is a person’s shelf, and pointing this at your own holdings is not the same as offering them up as somebody else’s search results. That holds for a Raster profile too: the artist page joins, the collection does not.
Three dollars in ETH on Base buys six credits. A look-up spends one, a curation spends two — a curation reads up to twenty-four pictures and then reasons over all of them, so it costs about twice what looking at one does. The credits are priced at what that actually costs at the API rather than marked up. No card and no account, because a collector here has always been allowed to be nobody in particular, and paying for a curation is not going to be the first thing that asks for a name.
What comes out is a proposal sheet, not a show. This gallery’s one rule is anonymity for the maker and credit for the eye — so the machine does the looking, and the curator who picks, cuts and hangs is the one whose name goes on the wall.
Everything hanging here, on Base.
A photograph of a painting, a screenshot, something you made this morning. It gets looked at the same way a minted work does, and held up against everything the tool has seen. Nothing needs to be on a chain, or finished.
It is not pinned, not stored, and never joins the pool other people search. It is resized in your browser first, so the camera, the timestamp and the location never leave your machine at all.
- A look-up — one picture, held against the pool
- 1
- A curation — reads the collection, proposes shows
- 2
A curation is worth two look-ups because it costs about that: a look-up reads one picture, a curation reads up to twenty-four and then reasons over all of them. Neither is a margin — the credits are priced at what the looking actually costs at the API.
Three dollars is the suggestion, not the price of entry. Send any amount and it converts at the same rate; anything left over stays on the balance and joins your next top-up rather than disappearing.
