the urufulabs studio ✿ how-to
eight little steps — drop your art, tune the weights, generate up to 10,000 tokens right in your browser, review and rework any of them, then publish to IPFS or download a zip you host yourself. no wallet needed to create ✿
how it works: urufulabs studio produces the whole collection combinatorially. you don't draw each token — you upload trait layers (background, body, eyes, hat…) and the studio combines them into thousands of unique tokens for you.
how big can it be? the math is trait count per layer, multiplied. 5×6 = 7,776 possible. 5×10 = 100,000. 6×8 = 262,144. want 10k? aim for at least ~2× headroom so the generator doesn't exhaust unique combinations.
click the ✿ untitled collection stamp at the top of the studio and give your set a name. this becomes the collection title in every token's metadata later ✿
drag a folder of subfolders into the pink dashed zone. each subfolder becomes a layer (background, body, eyes, hat…) and each png inside becomes a trait.
folder structure it expects:
my-collection/
├─ background/
│ ├─ sky.png
│ └─ dusk.png
├─ body/
│ ├─ pink.png
│ └─ mint.png
└─ hat/
├─ beanie.png
└─ crown.pngno folders? click + add layer / files instead and pick pngs one at a time.
you own everything you upload. click any layer or trait name to rename it — the pretty name shows up in the token metadata; the filename stays hidden.
not every trait should be equally common. bump the weight on a rare trait, drop it on a common one. the studio biases the roll toward higher weights when it generates.
use rules to keep pairings sensible:
scroll to generate collection ✿. set your target size (any number up to 10,000), hit the big button, and watch the counters climb.
during the run you get:
runs entirely in your browser. nothing uploads until you say so.
happy with the results? you have two ways out — pick either, or do both.
publish to ipfs — images get pinned as one directory, then a matching ERC-721 metadata directory referencing those images gets pinned right after. you get back one metadata CID your launchpad or contract uses as baseURI. token metadata lives at ipfs://METADATA_CID/{tokenId}.json.
download bundle — grab a single zip you can host anywhere (S3, arweave, your own gateway, whatever). inside you get images/, metadata/, and a README.txt explaining how to rewrite the image URL prefix after upload. set the prefix before you download if you already know where images are going, or leave the placeholder and find-and-replace later.
nothing leaves your browser until you press one of these buttons ｡◕‿‿◕｡
after generation, scroll to collection browser ✿. every finished token lives here — filter by trait, sort, or jump to a specific token id.
click any tile to open it. from there you can:
all edits happen locally. nothing re-uploads until you re-publish.
want a special token outside the algorithm? click Add 1-of-1 ✿ in the browser toolbar.
two ways to add one:
give it a name, a description, and any custom attributes you like. 1-of-1s append to the end (highest token ids) and survive regeneration — if you re-generate the algorithm collection later, your bonus tokens stay untouched ♡
frequently curious
no. every trait, weight, rule, and generated token lives in your browser storage until you press publish. we do not upload anything before then ✿
the target field caps at 10,000 to keep things smooth. below that, generate as many or as few as you like.
no wallet needed to design or generate. you only need one later when a launchpad or contract asks you to plug in the metadata CID.
nope. hit Download bundle ✿ instead and you get a zip with images/, metadata/, and a README. host it anywhere you like — s3, arweave, your own gateway. the README shows exactly how to swap the image URL prefix in every metadata json after you upload.
anything already rendered stays saved in your browser. cancel or refresh — you can resume with the same seed and pick up where you left off.
yes. layers, traits, rules, weights, and finished tokens persist between sessions on the same device.
no — the studio is generative. you upload trait layers once, and it combines them algorithmically into your whole collection. you don't hand-craft individual tokens.
yes. open the collection browser, click any token, and either reroll it (new random combo, same token id), or swap traits to hand-pick which trait from each layer it uses.
yes. in the collection browser, click Add 1-of-1 ✿. upload a fully-composed png or compose one from your traits. it appends to the end of the collection with a '1 of 1' marker and survives future full-collection regenerations.
when you press Generate again with 1-of-1s in the collection, the studio asks whether to keep them. keep = 1-of-1s untouched, algorithm tokens redone. wipe = everything cleared and restarted.
ready?