Question 1

What kinds of photos work?

Accepted Answer

Anything visual. Outfit photos work best when you want the piece-by-piece breakdown — the tool extracts garment types, silhouettes and era cues. But comic covers, magazine pages, anime / manga art, screenshots, product shots, even interior or food photography all return useful aesthetic classifications. For non-outfit images, the piece breakdown is skipped (since the 'clothes' aren't real garments), but the aesthetic, hashtags and marketplace context still apply.

Question 2

Is my image stored anywhere?

Accepted Answer

No. Images are sent to the classification service, processed in memory by the vision model, and discarded immediately after the response is returned. We don't keep copies, we don't train on your photos, and we don't share them. Your classification history (the reads you've saved) lives only in your browser's local storage — it never reaches our servers.

Question 3

How accurate is the classification?

Accepted Answer

The model uses a controlled vocabulary of ~350 named aesthetics plus freeform classification for anything outside that. Confidence scores reflect how well the visual cues match the aesthetic's signature traits. Above 80% confidence is typically very reliable; 60-80% is a defensible read with some ambiguity; below 60% the model is genuinely uncertain. Adjacent and supporting reads are deliberately less confident — they capture style influences without claiming the outfit IS that aesthetic.

Question 4

Can I use the suggested hashtags commercially?

Accepted Answer

Yes. The hashtags are derived from public aesthetic vocabulary and detected garment terms — there are no licensing restrictions. They're designed to be pasted directly into Depop, Grailed, Instagram, TikTok or eBay listings. Each platform has its own hashtag conventions; the suggestions are platform-agnostic but biased toward the resale and fashion communities.

Question 5

How fresh is the Grailed and Depop marketplace data?

Accepted Answer

Each aesthetic's marketplace data refreshes every 24 hours. The first time anyone classifies an aesthetic, the data is pulled live from Grailed's Algolia search and Depop's aggregates endpoint. Subsequent classifications of the same aesthetic within 24 hours use the cached snapshot for speed. The next day, fresh data populates on first request.

Question 6

Does this work for vintage or non-clothing items?

Accepted Answer

Yes. Vintage items still get aesthetic classification — a 90s graphic tee will read as Y2K or Streetwear, a 70s prairie dress as Cottagecore. Non-clothing items (decor, books, packaging, art) get classified by their visual aesthetic too. The piece-by-piece breakdown is only generated when the photo shows real wearable garments, but the aesthetic read and marketplace context apply to anything visual.

What aesthetic is this?

More tools

Aesthetic dictionary

Depop hashtag generator

Grailed hashtag generator

Depop fee calculator

Grailed fee calculator

All tools

Spend less time on admin. More time sourcing.