Isolate
Run a single effect — zoom, grain, pixel shift — without the rest of the stack, so the distance you see belongs to that effect alone.
Features
Every control, and what it actually does.
RepurposeThis is not a one-slider “uniqueizer”. It is a batch renderer with three separate pipelines — video, audio and container — and a measurement tool that tells you whether the output moved enough to matter.
Video Repurposer
Frame-level changes, randomised per copy.
Colour, geometry, grain and timing. You set the range for each effect; every copy samples a different value inside it, so no two outputs share the same frame fingerprint.
- Ranges, not values. Film noise 4–12, zoom 1.02–1.12, brightness −0.2–0.2, and so on.
- Overlays. A watermark or a background image, with position and opacity, baked into every copy.
- Retry reframe. Corner-anchored re-crop at a higher zoom when a copy needs a second pass.
- Audio-safe. Sources without an audio stream skip the audio graph instead of failing the encode.
Audio graph
Fifteen audio controls most tools skip.
Audio fingerprinting is often stricter than video hashing, because a re-encode keeps the waveform almost identical. These controls change the sample stream while staying inside “sounds the same to a person”.
| Control | What it changes | Range | Why it matters |
|---|---|---|---|
| Pitch | Semitone shift without changing tempo | −1 … +1 st | Moves the spectral fingerprint cheaply. |
| Waveform shift | Micro time-offset of the whole track | 1 … 20 ms | Inaudible, but it changes sample alignment. |
| Equalizer | Broad tonal tilt | −6 … +6 dB | Alters band energy without obvious colouration. |
| Reverb / Room echo | Short room tail | 40–120 ms | Breaks exact waveform matching. |
| Tremolo / Vibrato | Amplitude or pitch modulation | 2–8 Hz | Small, continuous variation across the file. |
| Chorus / Flanger / Phaser | Short modulated delays | 1–60 ms | Dense per-sample change, still musical. |
| Compression / De-esser / Limiter | Dynamics and peaks | −30 … −10 dB | Normalises the copy so it does not stand out. |
Fingerprint Lab
Measure before you render 200 copies.
Most tools give you a wall of sliders and no feedback. The lab runs one effect at a time on fast samples of your own source, measures how far the output moved, and ranks the effects that actually matter for that footage.
Rank
Effects are ordered by measured dHash distance for this source, not by folklore. A still portrait and a moving clip produce different rankings.
Apply the winners
“Apply top 3” enables only the effects that moved the fingerprint, then you can widen the ranges and run the real batch.
Image Uniquifier
Pixels and EXIF, handled separately.
Image uniqueness fails in two places: the pixels are too close to the source, or the metadata still names the original camera and edit time. RepurposeThis changes both, per file.
- 13 transforms: saturation, contrast, brightness, gamma, grain, vignette, zoom, rotation, pixel shift, crop, blurred border, flip, pattern overlay.
- Fresh EXIF with a plausible device, software version and capture time — never copied from the source.
- Format choice: keep the original, or force JPG, PNG, WebP or TIFF. Optional size reduction per format.
- Metadata stripped from the source before the new EXIF is written, so nothing leaks through.
Presets
Ten starting points, all editable.
A preset is a set of ranges plus flags, not a fixed recipe. Export one as .rt-preset.json and it behaves identically on another machine.
Built-in profiles
Presets tuned for short-form video, photo posts, aggressive movement and subtle delivery. Load one, then widen or narrow any range.
Save and share
Save the current stack under a name, edit it later, export it as JSON, import a file from someone else.
Per-source control
Different copy counts and presets per source in the same batch, so a folder does not have to be homogeneous.
Reliability
The parts that save a whole afternoon.
Hardware encoding with fallback
NVENC on NVIDIA, AMF on AMD, VideoToolbox on macOS. If the GPU encoder refuses a file, the pipeline retries on x264 instead of failing the batch. On a laptop with a hybrid GPU this is the difference between “done” and “Encoder failed (GPU + CPU)” at copy 300.
Cancel without leaving debris
Esc stops the run, and partial outputs are named so they are easy to spot and delete. Nothing is left half-written pretending to be finished.
Contact sheets
Build a grid montage of a batch with columns and filename labels. One image to review 50 copies before publishing any of them.
History you can read
Every run is logged locally with its preset, copy count and median distance — useful when a preset worked last month and you cannot remember its name.
Interface
Made to be used in three languages.
The interface ships in English, French and Arabic, with the whole layout flipped for right-to-left. Light and dark themes, and keyboard shortcuts that cover the whole workflow.
KeysGlobal
Shortcuts
Esc cancel · ? shortcut overlay · 1–7 switch workspace · Ctrl/Cmd + O browse.
ThemesTwo
Light and dark
Switch from the sidebar, not a settings page. The layout keeps the same contrast in both.
LayoutRTL
Arabic, properly
Not a translation string swap: the sidebar, the tables and the number fields mirror direction.
See it on your own footage.
Install the app, send us your machine ID, and test a preset on the source that matters to you.