Proprietary, on-device technology

The engines every clip runs on

BurnKit burns animated captions directly onto a video on the user's own phone, and ClipKit crops and pans a video into a highlight clip — both without ever uploading the video to a server. Real, shipped, running on every export today. Two years of studying how creators actually edit on their phones went into building them, grounded in our own background running content brands, not a feature we bolted on to look impressive.

The problem

Every caption meant an upload. We decided that was the wrong tradeoff.

Most caption and clip editing apps render server-side — which means your full, raw video has to leave your phone before a single word gets burned onto it.

Every export means an upload

A typical 10-minute video runs ~400MB. Uploading that just to burn on captions or crop a highlight clip is slow on real mobile networks, expensive to run at scale on the server side, and means a copy of your raw footage sits on someone else's infrastructure — a real privacy tradeoff most apps don't talk about.

The same 10 minutes of extracted audio is ~5-15MB. That gap is the entire reason BurnKit and ClipKit exist.

Off-the-shelf tooling wasn't enough

We tried the standard route first — the stock Android video-editing libraries most apps reach for. They broke down on real devices in ways that never showed up in a demo. Rather than ship something that only worked on the test phone, we built our own rendering pipeline from the ground up.

That decision is the entire reason BurnKit and ClipKit exist as real, hardened, in-house technology today — not a wrapper around someone else's SDK.

The solution

Two engines, built and hardened in-house

Not licensed, not a wrapper around someone else's SDK. Both live on Android today — BurnKit for iOS is coming soon.

BurnKit — on-device caption burn-in

Burns real, per-word-timed animated captions directly onto your video, on your phone, in real time. Supports all 6 of Cinematic Pro's rendering mechanics — recolor, phrase marker, box, word-box, glass-pill, and reveal-blur — including gradient and glow text and per-word cursive font swaps, pixel-matched against the server renderer.

  • All 38/38 caption templates ported and verified on real 4K device video
  • Per-word timing, not per-line — captions animate exactly on the beat
  • Zero video upload required to burn a caption

ClipKit — on-device face-tracked reframing

Turns a long landscape video into a vertical highlight clip entirely on the device — cropping and panning to keep the speaker's face in frame, using a real on-device face-tracking curve, then trimming and re-encoding without ever leaving the phone.

  • Real face-tracking, computed on-device
  • Built on the same in-house engine as BurnKit — one hardened technology, not two
  • A highlight clip never has to leave the phone just to be cropped and captioned
Why it's different

Built to run entirely on your phone

The details of the pipeline are ours — what matters to you is what it means in practice.

1

Your video never uploads

Captioning and clipping happen locally — no server round-trip, no full video sitting on someone else's infrastructure.

2

Full hardware quality

Runs on your device's real hardware video encoder — no compression shortcuts, no visible quality loss.

3

Real time, not "please wait"

Captions burn in and clips export in seconds, not minutes — see the exact numbers below.

4

Works offline

No network dependency for the render itself — the whole pipeline runs locally, start to finish.

The numbers

Not a marginal optimization

Measured on real devices, against our own real server-side renderer — not a synthetic benchmark.

0

real caption templates — every rendering mechanic — ported to BurnKit and visually verified on a real 4K device video.

0

faster caption burn-in on-device vs. our own server renderer — measured across 30 of 38 templates (~65s server-side vs. 2-3s on-device).

0

faster clip export with ClipKit — 356 seconds server-side vs. 15.86 seconds on-device, per clip.

Source: internal device benchmarks, Cinematic Pro engineering, 2026.

Why now

Ahead of a real industry shift

On-device AI processing isn't a niche bet — it's where the whole industry is moving, for exactly the reasons we built BurnKit and ClipKit.

Global edge AI market $30B → $118.7B
20262027202820292030203120322033
$118.7B
Projected global edge AI market by 2033
21.7%
Projected CAGR, 2026–2033
400MB → 15MB
Typical video vs. extracted audio, 10 minutes
38/38
Caption templates already running on-device

Source: Grand View Research, global edge AI market forecast, 2026. Privacy and low-latency processing are named as primary drivers of that shift — the exact case for BurnKit and ClipKit.

How we got here

Built from real experience running content, not a lab experiment

BurnKit and ClipKit weren't built by studying video editing in the abstract. They came out of more than two years spent studying how creators actually shoot, edit, and post — running our own content brands day to day, hitting the same slow-upload, laggy-preview, "just wait for the render" friction every creator on a phone deals with. That's the problem we set out to solve structurally, not paper over with a faster loading spinner.

Every engineering decision underneath BurnKit and ClipKit — including how the two share a common foundation instead of drifting apart as two separate technologies — was made deliberately, the same discipline you'd expect from infrastructure meant to be licensed and relied on, not a one-off feature.

FAQ

Common questions

What is on-device caption burning?

On-device caption burning means captions are rendered directly onto a video on the user's own phone, instead of uploading the video to a server to have captions added there. BurnKit is Cinematic Pro's own on-device caption-burning engine — it renders animated, per-word-timed captions locally, so the raw video never has to leave the device.

Why is on-device video processing better than server-side rendering?

On-device processing avoids uploading the full video file just to burn on a caption or crop a clip — a typical 10-minute video is around 400MB, versus 5-15MB for just the extracted audio. That means faster turnaround, lower infrastructure cost at scale, and no copy of raw footage sitting on a server.

How much faster is on-device caption rendering than server-side rendering?

BurnKit renders captions on-device in about 2-3 seconds, versus roughly 65 seconds for the same caption on Cinematic Pro's own server-side renderer — measured across 30 of 38 real caption styles on real devices, over 25x faster.

Can I license BurnKit or ClipKit for my own app?

Yes. BurnKit (on-device caption burning) and ClipKit (on-device face-tracked video reframing) are available for licensing to other products. Contact contact@weuinnovations.in with a licensing inquiry to request a quote.

What is ClipKit used for?

ClipKit is Cinematic Pro's on-device engine for turning a long landscape video into a vertical highlight clip — cropping and panning to keep the speaker in frame, then trimming and exporting, entirely on the device. It is roughly 20x faster than server-side clip export: about 15.86 seconds on-device versus 356 seconds server-side, per clip.

Is BurnKit available for iOS?

BurnKit and ClipKit currently run on Android. An iOS version of BurnKit is planned.

License BurnKit or ClipKit for your product

If your app renders captions or clips server-side today, BurnKit and ClipKit solve the exact bottleneck you're running into — available for licensing for other products.

Request a quote