Midjourney Showcase Data With Prompt Metadata
If you need to watch Midjourney's public showcase for emerging visual styles, study how creators phrase prompts, or pull style details into a research dataset, the slow part is rarely the idea.
It is the browser fleet, the proxy pool and the anti-bot arms race. A hosted scraping API turns each public Midjourney URL into a request: you send the URL and your key, the service renders the page and returns HTML or structured JSON. This guide explains what such a pipeline can collect, how to configure the first request, and where the real failure points sit.
What a Midjourney data pipeline can collect
The public Midjourney community feed is the usual starting point. Three workflows recur: watching style trends day over day, researching public prompts and their parameters, and extracting image style details for competitive or brand analysis. The fields that matter in practice are the image URL, the prompt text, style tags or modifiers, generation parameters, and the creator handle. A single request can return those fields as parsed HTML or as structured JSON when extraction rules are supplied. The limit is the public page itself: anything that renders on a public page is fair game for extraction, while post-login content is out of scope. Treat the feed as a moving target. Style tags and prompt conventions change, so a monitoring job should store the raw response alongside the parsed fields; that way a later schema change does not erase the evidence you collected earlier.
Request anatomy: URL, key and the parameters that change behaviour
The basic shape is a GET request to the scraping endpoint with the target Midjourney URL and your API key. From there, four parameters do most of the work. render_js tells the service to execute the page's JavaScript in a real headless browser and return the post-render HTML. stealth_proxy escalates to a stronger proxy path when a target is aggressive. country_code routes the request through a chosen geography. ai_extract_rules lets you describe the fields you want in natural language instead of writing selectors; extract_rules with CSS or XPath selectors is the deterministic alternative. A useful field/control map looks like this: input side — target URL, API key, render_js, stealth_proxy, country_code, extract_rules or ai_extract_rules; output side — image URL, prompt text, style tags, generation parameters, creator handle, plus the raw HTML if you ask for it. Keep the input and output sides separate in your code. Mixing them is the most common reason a pipeline that works in a notebook falls over in production.
A worked example: a daily style-trend job
Suppose you want to know which visual styles are rising in the community feed before they saturate. A simple daily job can work like this. First, collect a set of public Midjourney showcase URLs — for example, the community feed pages you already monitor. Second, send each URL with render_js enabled and an ai_extract_rules description asking for image URL, prompt text, style tags, generation parameters and creator handle. Third, store the JSON response with a timestamp and the source URL. Fourth, compute a daily count per style tag and compare it with the trailing window. The output is a ranked list of tags whose frequency is climbing. Two cautions make the difference between a useful signal and noise. Deduplicate by image URL, because the same showcase item can appear on more than one page. And keep the raw prompt text, not just the tags, because prompt phrasing often shifts before a tag becomes popular. If a run returns empty fields, check whether the page rendered at all before you change your extraction rules; a blank render and a wrong selector look identical in the output but have different fixes.
Three failure modes and how to tell them apart
Most failed Midjourney scrapes fall into one of three buckets, and the fix differs for each. The first is a JavaScript problem: the page loads but the data you want is injected after load. The symptom is an HTML response that contains the shell of the page but not the showcase items. The fix is to enable render_js so the page runs in a headless browser and you receive the post-render HTML. The second is IP-based blocking: requests are refused or served a challenge page because the source IP looks like a bot. The symptom is a consistent block across many URLs from the same worker. The fix is proxy rotation, which the service handles by default, with stealth_proxy as the escalation when a target is unusually aggressive. The third is a CAPTCHA or interactive challenge: the page renders but asks for human verification. The symptom is a challenge element in the HTML. The fix is the stealth path, not more retries on the same route. Keeping these three separate in your logs — render failure, block, challenge — turns a vague 'it stopped working' into an actionable alert.
Setup path and language coverage
The setup sequence is short. Create an account to get an API key and test credits. Install the client for your stack, or skip the SDK and use raw cURL. Send a request with the target URL and your key. The service renders the page, rotates proxies and returns the data. Client libraries are listed for Python, Node.js, Go, PHP, Ruby and Java, with raw cURL examples alongside. The dashboard also includes a request builder: paste a URL, set parameters, and copy the generated snippet in Python, Node, Go, PHP, Ruby or cURL. That builder is the practical answer for readers who want to test a configuration before writing code. For teams already using workflow tools, an n8n integration can run prompt-tracking and style-monitoring jobs without scraper code. A reasonable first milestone is one successful request against a single public showcase URL, with the JSON saved to disk. Only after that should you wire the job into a scheduler.
Pricing, credits and what to check before you commit
Plan prices, credit allowances and concurrency limits change, so treat any figure you see on a marketing page as a dated reference and confirm it on the current pricing page before you budget. What stays stable is the credit logic. API credits are valid for one month and do not roll over. Requests start at one credit each. Enabling JavaScript rendering costs five API credits per successful request. Premium proxies cost ten or twenty-five API credits per successful request, depending on whether JavaScript rendering is also enabled. That means a Midjourney job with rendering plus premium proxies is materially more expensive per page than a plain HTML fetch, and the difference should be modelled before you scale. A practical checklist: estimate pages per day, multiply by the credit cost of your actual parameter combination, add a retry margin, and compare the result with the plan's monthly credit allowance. If the estimate is close to the limit, choose a plan with headroom rather than one that fits exactly.
Legal scope and responsible use
Public-facing pages can usually be scraped, but the target site's terms of service should be reviewed before deployment, and post-login content should be avoided. That is the operating boundary for a Midjourney pipeline: public showcase pages, public prompts, public style metadata. Anything behind an account is outside the scope of this workflow. The service states that it does not collect or store personal data from scraped sites unless the user explicitly requests it, and it describes itself as GDPR and CCPA compliant; those are the provider's stated positions and should be verified against the current legal pages for your jurisdiction and use case. A short internal checklist helps: confirm the pages are public, confirm the terms permit the intended use, confirm no personal data is being pulled in, and record the date of that review so the decision can be revisited when terms change.
| Plan | Monthly price | API credits | Concurrent requests | JavaScript rendering | Rotating & premium proxies | Geotargeting | Screenshots & extraction rules | Priority email support | Dedicated account manager | Team management |
|---|---|---|---|---|---|---|---|---|---|---|
| Hobby | $19 /mo | 75,000 | 25 | Included | Included | Included | Included | Not included | Not included | Not included |
| Freelance | $49 /mo | 250,000 | 50 | Included | Included | Included | Included | Not included | Not included | Not included |
| Startup (Recommended) | $99 /mo | 1,000,000 | 100 | Included | Included | Included | Included | Not included | Not included | Not included |
| Business | $249 /mo | 3,000,000 | 200 | Included | Included | Included | Included | Included | Included | Included |
| Business + | $599 /mo | 8,000,000 | 400 | Included | Included | Included | Included | Included | Included | Included |
| Enterprise 14 | $999 /mo | 14,000,000 | 500 | Included | Included | Included | Included | Included | Included | Included |
| Enterprise 24 | $1,599 /mo | 24,000,000 | 600 | Included | Included | Included | Included | Included | Included | Included |
| Enterprise 41 (Recommended) | $2,399 /mo | 41,000,000 | 700 | Included | Included | Included | Included | Included | Included | Included |
| Enterprise 69 | $3,599 /mo | 69,000,000 | 800 | Included | Included | Included | Included | Included | Included | Included |
| Enterprise 120 | $5,799 /mo | 120,000,000 | 900 | Included | Included | Included | Included | Included | Included | Included |
Plan prices, credit allowances and concurrency limits are a dated reference captured from the provider's pricing page. All prices are exclusive of VAT. API credits are valid for one month and do not roll over. Confirm current figures on the live pricing page before budgeting.
| Configuration | Credits per successful request |
|---|---|
| Plain request (no JavaScript rendering, no premium proxies) | 1 credit (requests start at 1 credit each) |
| JavaScript rendering enabled | 5 credits |
| Premium proxies, JavaScript rendering disabled | 10 credits |
| Premium proxies, JavaScript rendering enabled | 25 credits |
Credit costs apply to successful requests. API credits are valid for one month and do not roll over. Confirm current credit rules on the pricing page.
Frequently asked questions
Is scraping Midjourney legal?
Public-facing pages can usually be scraped, but review the target site's terms of service before deploying and avoid post-login content. The provider states that it does not collect or store personal data from scraped sites unless the user explicitly requests it, and describes itself as GDPR and CCPA compliant; verify those positions against current legal pages for your jurisdiction and use case.
What data can I extract from Midjourney?
The practical boundary is what the public page actually displays: if a visitor can see it without signing in, it can be targeted. To turn that into structured JSON, you have two routes. Selector-based extract_rules pin down specific CSS or XPath elements and give you a predictable, repeatable shape. AI-driven ai_extract_rules instead take a plain-language description of the fields you want and let the service map them, which tolerates layout drift but should be spot-checked against a sample before a production run.
How do I start scraping Midjourney?
Sign up, get your API key, and send a GET request with the target URL. The service renders the page, rotates proxies and returns the data. Test credits are available to validate the first request before you commit to a plan.
Will I get blocked when scraping Midjourney?
The service rotates residential proxies and uses a real headless browser so requests resemble normal traffic. If a target becomes aggressive, enable stealth_proxy on the request. Persistent blocks across many URLs from one worker usually indicate an IP-level issue rather than a rendering issue.
Can I scrape Midjourney without writing code?
Yes. The dashboard request builder lets you paste a URL, set parameters, and copy the generated snippet in Python, Node, Go, PHP, Ruby or cURL. For workflow automation, an n8n integration can run prompt-tracking and style-monitoring jobs without scraper code.
How do I handle JavaScript when scraping Midjourney?
Pass render_js=true on the request. The page runs in a real headless browser, the JavaScript executes server-side, and you receive the post-render HTML. If the response contains the page shell but not the showcase items, rendering is the first thing to check.
How much does a Midjourney scrape cost in credits?
Requests start at one credit each. JavaScript rendering costs five credits per successful request. Premium proxies cost ten or twenty-five credits per successful request depending on whether rendering is also enabled. Credits are valid for one month and do not roll over, so model your daily page count against the monthly allowance before choosing a plan.
What is the difference between extract_rules and ai_extract_rules?
extract_rules uses CSS or XPath selectors and is deterministic: you specify exactly which elements to pull. ai_extract_rules describes the fields you want in natural language and lets the service map them. Selector-based rules are better for stable page structures; AI rules are more tolerant of layout changes but should be validated against a sample before a production run.