Japan Used Camera Market — Multi-shop Search, Stats & Alerts

Quick start

1. Click Try for free / open the Actor in Apify Console.
2. Keep the prefilled input ({"mode": "search", "sources": ["fujiya", "sanpou", "naniwa", "hardoff"], "keywords": ["Leica"], "maxItems": 10}) or type your own keywords.
3. Press Start — listings from all four shops appear in the Output tab in about 20–30 seconds. (Without maxItems the default is 200 matches, which can mean reading most of the shops' listing pages, a minute or two.)

Search, benchmark and monitor Japan's used camera shops — second-hand cameras and lenses from Tokyo's specialist shops — with one unified schema. Built by someone who reads Japanese and knows what AB- vs 現状品 actually means.

Japan is the world's deepest used-camera market (Leica, Voigtländer, Contax, Nikon F, Hasselblad, Fujifilm…), but the shops only publish in Japanese, each with its own condition grading. This Actor turns them into clean JSON/CSV you can pipe into Google Sheets, n8n, Slack, Discord or your own pricing model. Use it to buy a used camera from Japan online, to check the Japan price of a Leica Q3, M11 or M6 before you order, or to source used Japanese cameras for resale.

What you can do

Mode	What it does	Who uses it
search	Pull listings matching keywords/brand/condition grade/price. Sold-out listings are excluded unless includeSoldOut: true.	eBay "From JAPAN" sellers sourcing stock, collectors hunting a specific body, proxy-buying services
stats	Model stats of in-stock listings: count, min / median / max in JPY and USD, broken down by shop and condition grade, plus the cheapest listing URL.	KEH / MPB / regional dealers benchmarking against Tokyo prices, anyone pricing a trade-in
monitor	Run on a schedule. Keeps a snapshot and emits only what changed: NEW, PRICE_DROP, PRICE_UP, SOLD_OUT (sold-out listings are tracked internally regardless of includeSoldOut; maxItems is ignored). Optional webhook → Slack / Discord / n8n / Make / Zapier.	Anyone who wants to be first on a one-of-a-kind listing the minute it's listed
history	Daily price history per model: min / median / count for each day, by shop and condition grade, from stored daily snapshots — no crawling. Needs keywords, brands or models. History is still being collected (since 2026-11-05), so treat the first 30 days as indicative.	Anyone timing a purchase or sale, dealers tracking how Tokyo prices move

Shops

Shop	Status	Listings	Notes
Fujiya Camera (fujiya, フジヤカメラ, Nakano Tokyo)	✅ live	~3,500	full used stock; grade, brand, price, stock status (incl. 売り切れ), image
Sanpou Camera (sanpou, 三宝カメラ, Jiyugaoka Tokyo)	✅ live	~1,300	full used stock (cameras, lenses, accessories); grade, brand, price, stock status (incl. 売り切れ), image
Naniwa / Lemon-sha / Takachiho (naniwa, カメラのナニワ・レモン社)	✅ live — new arrivals only (~200–300)	~200–300	the shop's full-stock search is closed to crawlers (robots.txt), so only its "new used arrivals" list is read. Great for monitor mode; not a full price benchmark
Hard Off Off Mall (hardoff, ハードオフ オフモール, nationwide stores)	✅ live	~16,000	cameras and lenses on sale ("販売中のみ"), newest first; Hard Off's own N / S / A / B / C / D / ジャンク ranks. Sold listings drop out of the list, so a Hard Off SOLD_OUT is detected on full scans. In the price history: weekly (Sundays)
Kitamura (ネット中古)	✗ not crawlable		listings are rendered by JavaScript from a private, key-protected API
Kitamura Camera Shinjuku (北村写真機店)	✗ no online listing		its ~5,000 used items are sold in store / via Kitamura's site

Want a shop added? Open an issue on the Actor page — if it's publicly crawlable it goes on the roadmap.

Condition grades, normalized

Every shop's label is mapped to one scale so you can compare across shops:

NEW > A > AB+ > AB > AB- > B > ASIS (sold as-is, no warranty) > PARTS (parts / collection only)

The raw label is always kept in conditionGradeRaw. Shops use the same letters differently, so each is mapped with its own published scale:

Shop	Raw labels → normalized
Fujiya	未使用/未開封 → NEW · A · AB+ · AB · AB- · B · 現状品 → ASIS · コレクション/ジャンク → PARTS
Sanpou	未使用品, 逆輸入品 → NEW · S → A · A → AB+ · AB → AB · B → B · C (major parts missing) → ASIS
Naniwa	新同 → NEW · A → A · AB+ → AB+ · AB → AB · B+ → AB- · B, B- → B · C (some faults) → ASIS · ジャンク → PARTS
Hard Off	N (new, sealed) → NEW · S (like new) → A · A (barely used) → AB+ · B (normal use, few marks) → AB · C (visible use, many marks) → AB- · D (heavy wear) → B · ジャンク → PARTS

Input

{
  "mode": "search",
  "sources": ["fujiya", "sanpou", "naniwa"],
  "keywords": ["Leica M", "Noctilux"],
  "grades": ["A", "AB+", "AB"],
  "maxPriceJpy": 600000,
  "maxItems": 100
}

Monitor example (schedule it hourly):

{
  "mode": "monitor",
  "sources": ["fujiya", "sanpou", "naniwa"],
  "keywords": ["Voigtlander", "Leica"],
  "alertWebhookUrl": "https://hooks.slack.com/services/XXX/YYY/ZZZ",
  "alertEvents": ["NEW", "PRICE_DROP"]
}

The first monitor run only builds the snapshot (full scan) and emits nothing; every run after that emits monitor events. Only the event types listed in alertEvents are written to the dataset, charged and sent to the webhook.

Snapshots per shop and filter. Each monitor keeps one snapshot per shop, keyed by its filters (keywords, brands, grades, minPriceJpy, maxPriceJpy, includeSoldOut), so schedules with different filters never interfere. If you run several schedules with the same filters, give each a distinct monitorId.

Request budget. Every request is one listing page (Fujiya 50 listings, Sanpou 28, Naniwa ~25, Hard Off 30), read newest-first; product pages are never opened. By default monitor stops paging once it hits 20 already-known listings in a row in each list — an hourly monitor on all four shops is normally 7 requests (Fujiya 1, Sanpou 3 — cameras, lenses, accessories — Naniwa 1, Hard Off 2 — cameras, lenses). That catches every NEW listing and any sell-out or price change on those first pages, but changes to older listings further down can be missed — set "fullScan": true to read all pages (full scan ≈ 670 requests, ~25 minutes: Fujiya ≈ 70, Sanpou ≈ 50, Naniwa ≈ 8, Hard Off ≈ 540 — leave Hard Off out of sources for a ~4-minute full scan; and the only mode in which a listing that vanished from the listing pages counts as SOLD_OUT). A good setup is a frequent light monitor plus one daily full scan.

Measured run times (Fujiya, direct connection, Nov 2026): run time grows with how many listing pages it takes to find maxItems matches — search "Leica", 20 listings ≈ 11 pages ≈ 40 s; no keyword filter, 20 listings ≈ 1 page; full scan ≈ 65 s.

Output schema

{
  "source": "fujiya",
  "sourceName": "Fujiya Camera",
  "sourceItemId": "C2120187965805",
  "url": "https://www.fujiya-camera.co.jp/shop/g/gC2120187965805/",
  "imageUrl": "https://…/C2120187965805.jpg",
  "title": "FUJIFILM X-T20 ボディ ブラック",
  "brand": "FUJIFILM",
  "brandJa": "フジフイルム",
  "model": "fujifilm x-t20",
  "variant": null,
  "conditionGrade": "AB-",
  "conditionGradeRaw": "AB-",
  "priceJpy": 92950,
  "priceUsd": 619.67,
  "fxJpyPerUsd": 150,
  "inStock": true,
  "category": "used",
  "scrapedAt": "2026-11-04T00:00:00.000Z"
}

Monitor events add eventType; PRICE_DROP / PRICE_UP add previousPriceJpy, priceChangeJpy, priceChangePct, and a SOLD_OUT for a listing that flipped to 売り切れ adds previousPriceJpy. A SOLD_OUT for a listing that disappeared from the listing pages (full scan only) carries the last known fields from the snapshot. Listings first seen already sold out are not reported as NEW.

There are no serial, listedAt or condition-note fields: those live on product pages, which this Actor does not open (see Limitations).

Model stats rows look like:

{
  "model": "leica m6",
  "brand": "Leica",
  "sampleTitle": "M6 ボディ ブラック",
  "count": 7,
  "variants": [],
  "minJpy": 398000, "medianJpy": 455000, "maxJpy": 528000,
  "minUsd": 2653.33, "medianUsd": 3033.33,
  "cheapestUrl": "https://…",
  "cheapestGrade": "AB-",
  "byGrade": { "A": { "count": 2, "minJpy": 498000, "medianJpy": 513000, "maxJpy": 528000 }, "AB": { … } },
  "byShop":  { "fujiya": { "count": 7, … } },
  "fxJpyPerUsd": 150,
  "computedAt": "2026-11-04T00:00:00.000Z"
}

Stats only count in-stock listings with a price; includeSoldOut has no effect in stats mode.

Price history since 2026-11-05

The Actor's own scheduled full scans store the complete inventory of each shop (every listing, price, grade and stock flag): Fujiya, Sanpou and Naniwa every day, and Hard Off once a week, on Sundays (its ~16,000 listings take ~540 pages, about 17 minutes) as one compressed snapshot per shop per day. It keeps them indefinitely. history mode reads those snapshots, so it costs no shop requests and returns in seconds:

{ "mode": "history", "keywords": ["Leica M6"], "days": 90 }

or by exact model key (as shown in the model field of stats mode):

{ "mode": "history", "models": ["fujifilm x100vi", "nikon zf"], "days": 30 }

One row per model × day:

{
  "model": "leica m6", "date": "2026-11-20", "brand": "Leica",
  "count": 3, "minJpy": 390000, "medianJpy": 430000, "maxJpy": 460000,
  "byShop":  { "fujiya": { "count": 2, "minJpy": 390000, "medianJpy": 425000, "maxJpy": 460000 }, "sanpou": { "count": 1, … } },
  "byGrade": { "A": { … }, "AB+": { … }, "AB": { … } },
  "byShopGrade": { "fujiya": { "A": { … }, "AB": { … } }, "sanpou": { "AB+": { … } } }
}

• Only in-stock listings with a price count. Days on which a model wasn't listed have no row.
• Hard Off is weekly: its numbers appear only on Sunday rows (byShop.hardoff). On the other days a row covers Fujiya, Sanpou and Naniwa only, so compare Hard Off week to week, not day to day.
• History is still being collected. Snapshots start on 2026-11-05, so for the first 30 days the series is short and should be treated as indicative only.
• Naniwa contributes its new-arrivals list only (see Shops), so Naniwa numbers describe what was newly listed, not its full stock.
• Model keys are recomputed when you query, so grouping improvements apply to old days too.

Pricing

Pay per event, no subscription:

Event	Price
listing (search mode, per listing)	$0.003
model-stats (stats mode, per model row)	$0.005
alert-event (monitor mode, per monitor event: NEW / PRICE_DROP / PRICE_UP / SOLD_OUT)	$0.01
history-row (history mode, per model × day row)	see the Actor's Pricing tab

A monitor run that finds nothing new costs you only platform compute (~$0.01–0.03). Apify's free plan includes $5/month of usage, enough for hourly monitoring of a keyword set.

Integrations

• Google Sheets: Apify → Integrations → Google Sheets, pick the dataset.
• Slack / Discord: paste the incoming-webhook URL into alertWebhookUrl; messages are pre-formatted.
• n8n / Make / Zapier: use the Apify node with this Actor, or point alertWebhookUrl at your workflow's webhook trigger.
• AI agents: see the next section.

AI agents (MCP)

A dedicated MCP server, @keiya-dev/japan-used-market-mcp (open source, MIT), gives Claude, Cursor and other MCP clients tools backed by this Actor. The same server also covers used watches ("market": "watch"); cameras are the default, so no market is needed here.

Tool	Mode
search_listings	search
price_stats	stats
price_history	history
start_monitor	creates a scheduled monitor (Task + Schedule) in your Apify account
list_markets	shops and grades per market (no run)

Claude Desktop: add this to claude_desktop_config.json. On macOS it is ~/Library/Application Support/Claude/claude_desktop_config.json; on Windows, %APPDATA%\Claude\claude_desktop_config.json. Then restart the app:

{
  "mcpServers": {
    "japan-used-market": {
      "command": "npx",
      "args": ["-y", "@keiya-dev/japan-used-market-mcp"],
      "env": { "APIFY_TOKEN": "apify_api_..." }
    }
  }
}

Get the token at Apify Console → Settings → API & Integrations. Runs are billed to your Apify account, as with any other run.

Then just ask: "What's the cheapest AB-or-better Leica M6 in Tokyo shops right now?", "Compare Nikon Zf prices by shop and grade", "How has the median X100VI price moved over the last 30 days?", "Watch for new Voigtländer lenses under ¥80,000 and post them to my Slack webhook."

Claude Code: claude mcp add japan-used-market -e APIFY_TOKEN=apify_api_... -- npx -y @keiya-dev/japan-used-market-mcp

Remote / HTTP clients: run npx -y @keiya-dev/japan-used-market-mcp --http --port 3000. The endpoint is /mcp (Streamable HTTP); send your token as Authorization: Bearer <APIFY_TOKEN>.

Already using the old @keiya-dev/japan-used-camera-mcp? It keeps working (it now starts the new server); switch the package name when convenient.

The generic Apify MCP server also works: add the Actor keiya_dev/japan-used-camera-market as a tool.

Limitations & fair use

• Prices are tax-included retail (税込) in JPY; USD is a live mid-market conversion (frankfurter.app) for convenience, not a quote. If that FX API is unreachable, the run still succeeds and uses a fallback rate of 150 JPY per USD (logged as a warning, visible in fxJpyPerUsd); set jpyPerUsd to use your own rate.
• Naniwa is new arrivals only (~200–300 listings at a time), so a model missing from Naniwa in search/stats doesn't mean Naniwa doesn't have it.
• Hard Off is read with its "on sale only" filter (its sort and search parameters are closed to crawlers by robots.txt, its default order is newest first), so Hard Off listings are always in stock and a sold one simply disappears.
• Fujiya, Sanpou and Naniwa keep sold listings on their listing pages for a while (marked 売り切れ / SOLD OUT, Sanpou without a price); inStock is read from that badge. SOLD_OUT fires when a listing flips to sold out on a page the run reads, or, on a full scan only, when it disappears from the listing pages.
• Serial numbers, listing dates and seller notes live on product pages, which this Actor does not open (Fujiya serves them behind a bot-protection JS challenge). Everything comes from the listing pages. Use the time of the NEW event as the listing date.
• Crawling is polite (concurrency 2, 300–700 ms random delay, no login, public listing pages only). If the shop's bot protection challenges a run, a listing page can't be loaded, or the shop changes its page layout, the run fails with a message saying what happened and what to do next — without writing partial data, charging, or updating the monitor snapshot. Please don't hammer monitor mode more than every 15 minutes — nothing changes faster than that.
• This Actor is not affiliated with any of the shops.

Changelog

• 0.4 — Fourth shop: Hard Off Off Mall (cameras + lenses on sale, ~16,000 listings, its N–D / ジャンク ranks mapped to the common scale). Model keys keep the model code when a title is nothing but a code (e.g. "WG-30").
• 0.3 — history mode: daily price history per model by shop and grade, from daily full-inventory snapshots (collected since 2026-11-05).
• 0.2 — Two more shops: Sanpou Camera (full stock) and Naniwa / Lemon-sha / Takachiho (new arrivals). Per-shop condition-grade scales. All three shops are selected by default.
• 0.1 — Fujiya Camera: search / stats / monitor, webhook for monitor events.

Related

• Japan Used Watch Market: the same four modes for used watches (WATCHNIAN / Ippuukishi, Hard Off Off Mall), with reference numbers and box / warranty card. Grand Seiko, Rolex, Omega and more.

---

If this saved you a trip to Nakano, a review on the Store helps others find it.