Introduction

FastDrop makes user-uploaded video safe to build on. Send a video URL. Get back a verdict — proceed, fix first, or reject — with exact instructions for anything that needs fixing. One request, answered in the same call, 1 credit.

What goes wrong without it

Footage that plays fine in a browser can still be wrong for what you are about to do with it. These defects are invisible on playback and surface only once your feature is in front of users:

Captions that start in sync and have drifted a full second by minute three
Clips cut on timestamps that land a beat late, beheading the punchline
A scoring or tracking feature that returns the wrong number — no crash, just a result that tells a user who did everything right that they did not
Two camera angles that slowly slide apart across a long take
The same clip stored, converted and analyzed four times, because nothing noticed it had already been through

None of these announce themselves. The user will not suspect the file — it played fine when they uploaded it — so the blame lands on your app.

Why timing problems hide during testing

Some files record each frame with a slightly different gap before the next one. Every gap is off by a tiny amount, and the errors accumulate: invisible at second one, a blink at second thirty, a full second by minute five.Short demo clips are too short for the error to add up, so this survives testing intact and appears the first time someone uploads a real video. Anything that reads a single average frame rate off the file reports a clean number and moves on.

Start here

Check the footage before committing to work on it:

curl -X POST https://api.fastdrop.io/api/v1/readiness \
  -H "Content-Type: application/json" \
  -H "X-API-Key: fd_live_your_key_here" \
  -d '{
    "video_url": "https://example.com/upload.mp4",
    "context": "nle_editing"
  }'

Response

{
  "verdict": {
    "action": "fix_first",
    "context": "nle_editing@v1",
    "reasons": [
      {
        "check": "vfr",
        "severity": "high",
        "detail": "Variable frame rate: the gaps between frames are uneven, so playback time and frame count disagree.",
        "fix": {
          "operation": "transcode",
          "to": { "frame_rate_mode": "constant", "fps": 29.97 },
          "reason": "Constant frame rate keeps audio and video aligned."
        }
      }
    ],
    "user_message": "Fix this before you start."
  },
  "credits_charged": 1
}

Three verdicts, and you branch on action:

`action`	What it means	What you do
`proceed`	Nothing found that blocks this use	Go ahead
`fix_first`	Something is wrong and can be fixed	Fix it yourself, or have us do it
`reject`	The footage cannot serve this purpose, and no transformation helps	Do not spend the rest

The fix is machine-readable, so you can execute it on your own infrastructure and the media never leaves it. Or hand it back: POST /fix runs exactly that instruction and then re-checks the result before returning it — a file comes back only if the original failure is confirmed gone. If that cannot be shown, you get the reason and are charged nothing.

Readiness is not a quality score. The same file is fit for one job and unfit for another, so what you are about to do with it is part of the question — pass a context. One clip can be fix_first for an editor, proceed for a social feed and reject for motion tracking, with nothing about the file changing between those calls. Why check footage before you process it walks through it.

Who this is for

FastDrop is for applications that measure video — caption it, cut it, score it, sync it, classify it, search it. Anything that derives a result from the content or the timeline is exposed to the defects above, because the measurement inherits every error in the file. If your app only stores and plays video back, you do not need a preflight check, and we would rather say so than sell you one. Players are forgiving in exactly the ways measurement is not. Worth knowing where the line is, though: the roadmap of most video products eventually grows a feature that measures something.

What you can do once footage passes

Every capability runs independently — request one, or several in a single call:

Classify — assign one of six editor-ready roles with confidence, explanation, suggested filename and folder path (2 credits)
Thumbnails — 3-5 keyframes per video (1 credit)
Diagnostics — the raw measurements behind a verdict: codec, resolution, frame rate, health score (1 credit)
Duplicates — perceptual fingerprinting, so you stop paying to process the same clip twice (1 credit)
Clips — clip candidate timestamps with labels (2 credits)
Transcribe — speech transcription in 50+ languages, English translation, optional SRT/TXT (8 credits)

See What you can ask for, and what it costs for the full breakdown, including the six classification roles and batch processing.

How results reach you

Readiness answers inline, in the request. A verdict you have to wait for arrives after the moment you needed it, so it does not hand you a job. Everything else is real work measured in minutes, and returns a job:

You submit a public video URL
FastDrop downloads and analyzes it asynchronously
The result is pushed to your webhook, or returned by holding a status request open with ?wait=N if your service cannot receive callbacks

Processing takes 30-180 seconds depending on video length and tier.

You never need a polling loop. Register a webhook and results arrive the moment they exist; otherwise ?wait=60 returns the finished job in a single request. Getting your results covers when to use which.

Integration options

REST API

Direct HTTP calls from any language. Full control over the integration.

MCP Server

Connect AI agents (Claude, Cursor, VS Code) to FastDrop via natural language.

Pricing

FastDrop uses a credit-based system. Capabilities cost credits per video processed; checking readiness first is the cheapest call in the API.

Plan	Price	Credits/month
Free	$0	100
Starter	$9/mo	150
Growth	$49/mo	1,000
Scale	$179/mo	5,000

A readiness check costs 1 credit, against 10 for a classify-and-transcribe pipeline on the same video — or 19 to run every capability at the premium tier. The check pays for itself well before it catches its tenth bad clip, and the clips it catches are the ones that would have burned the full run.

Get Started

Understanding video defects

Guides

What goes wrong without it

Start here

Who this is for

What you can do once footage passes

How results reach you

Integration options

REST API

MCP Server

Pricing

Get your API key

​What goes wrong without it

​Start here

​Who this is for

​What you can do once footage passes

​How results reach you

​Integration options

REST API

MCP Server

​Pricing

Get your API key

What goes wrong without it

Start here

Who this is for

What you can do once footage passes

How results reach you

Integration options

Pricing