1 video.100 uploads the platform reads as original.
VideoForge re-forges a single source video into dozens of algorithmically distinct variants, verified against Meta's PDQ perceptual hash and audio fingerprinting, so every account uploads something the platform treats as original.
Post the same file twice and the platform already knows.
Meta, TikTok, and YouTube run perceptual hashing on every upload. The same master posted across a fleet resolves to one fingerprint, gets matched as a re-upload, and quietly loses reach. Identical bytes are a liability at scale.
The frame stays watchable. The fingerprint does not. That gap, real distinctness verified against PDQ and audio fingerprinting, is the whole job.
One pipeline, tuned export profiles for the surfaces you actually post to. Correct aspect ratio, bitrate, and container per platform, every time.
Drop your master.
One source video in. MP4 or MOV, vertical or landscape. It stays on your machine until you forge.
Five layers, run in this order. Nothing hidden.
This is the actual pipeline sequence, not a marketing simplification. Each layer targets a specific axis a platform hasher checks, structure, composition, overlay, audio, before anything gets measured.
Zoom, flip, border, graph shuffle
Intensity-scaled zoom-crop, a first-class flip, a micro-border, and a per-job filter-graph shuffle. The moves that shift PDQ and TMK, not a cosmetic color pass.
Per-variant rect + zoom jitter
An off-center punch-in plus a gentle animated pan on every variant, so composition drifts frame to frame without ever reading as distorted.
Caption reblocking
Per-variant hook text, dealt round-robin and redrawn inside a rotating platform-safe zone, so no two exports carry a caption in the same spot.
Pitch, atempo, decorrelated bed
Real pitch and tempo shift on the source track, plus a quiet synthesized noise bed mixed underneath, built to move Chromaprint-class fingerprints without touching the mix you hear.
PDQ + Chromaprint metascoring
Every output is re-hashed and scored against the actual platform matchers, Meta's PDQ perceptual hash for video and Chromaprint for audio, measured on the bytes you download, not a self-reported figure. The report is the product.
One frame, two realities.
Drag the divider. What you see barely moves. What the platform hashes does.
Same source frame, measured pixel-for-pixel and hash-for-hash against its forged variant.
Visually identical. Algorithmically distinct.
// sample run, shown for illustration. your numbers will vary by source and intensity.
One upload becomes a verified fleet.
Built for operators posting the same creative across many accounts, not one clip at a time. Queue a batch, forge every variant, verify every output before it ships.
A fleet doesn't need one clean export, it needs a distinct one for every account it lands on. VideoForge fans a single source out into a batch, runs the same pipeline on each variant, and checks the whole set, not just against the source, but against each other, before you route anything downstream.
Most tools change how a file looks. Few change how it hashes.
A visual filter is not a uniqueness strategy. This is a capability comparison, not a scorecard against any one product.
Ground-truth PDQ + Chromaprint verification
Every variant re-hashed and scored against the source, not just visually eyeballed.
Audio decorrelation
Pitch, timing, and fingerprint shifted so the audio hash moves too.
Per-variant visual jitter
Crop, zoom, and color drift that differs on every output, not one global filter.
Batch / fleet output
One upload becomes N distinct variants without N manual passes.
Uniqueness report per variant
A readable PDQ Δ, SSIM, and audio delta for each file you ship.
Metadata stripping + re-encode
Source container fingerprints and embedded metadata removed, not just re-wrapped.
Pick your throughput.
Every plan runs the same pipeline end to end: visual transforms, audio decorrelation, PDQ frame distance and Chromaprint verification on every output. Higher tiers unlock more verified variants per month, more concurrency, and API access.
Previews, retunes, and verification passes are always free. A credit is spent only when a variant renders, passes verification, and gets exported.
Every number on the cards below is the real, public grant, not a teaser that quietly tightens after signup.
Free
For trying it out
- Full uniquification pipeline
- Full PDQ + Chromaprint verification report
- 1 video at a time
- Per-account history
Starter
For a first fleet outgrowing Free
- All uniquification recipes
- Per-variant verification scores
- Per-platform export profiles
- Full job history
Pro
For active creators running a fleet
- Priority processing
- Exportable audit reports
- Sibling-distinctness checks on every batch
- REST API + MCP access (Claude, Cursor)
Agency
For teams and agencies
- Fleet dashboard (variant → account)
- Shared brand-kit presets
- Outbound webhooks
- Priority queue + private Slack support
Enterprise
talk to usFor high-volume, dedicated throughput
Curious what a credit really costs us to fulfill, compute, storage, and egress, not just what it costs you? Read the honest breakdown →
Cancel any time from Billing. Your plan stays active through the end of the current period.
Straight answers.
If a claim about detection sounds too good to be true, it isn't from us. Here's what the pipeline actually does, in plain terms.
No tool can honestly promise that, and we won't claim it. What VideoForge does is engineer real, measurable distinctness into every variant, geometric jitter, per-clip content zoom, caption reblocking, audio pitch and decorrelation, then verify the result against PDQ frame distance (Meta's own perceptual hash) and Chromaprint audio fingerprinting before you download it. You get a verification report, not a marketing promise.
Stop uploading the same file everywhere.
Drop in one source clip, forge as many verified-distinct variants as your plan allows. No watermark, no card required.
PDQ-verified · audio-decorrelated · no watermarks