Send one master file. Forge reads what it really is, builds the right set of renditions for it, encodes them and packages a stream your players can switch through cleanly — on the same platform that marks your ad breaks and inserts your ads.
Nothing inside a general-purpose encoder knows what an ad break is, what an SSAI slot is or what a campaign is. The file comes back correct and still unsellable — someone has to carry it to a second system before it earns anything.
Forge is already inside that second system. The breaks, the overlays and the insertion sit on the same platform the file was encoded on.
Six stages, each one visible while it runs. No black box between upload and download.
| Stage | What it does |
|---|---|
| Ingest | Upload a master file into a workspace; the job is queued, not run on your machine |
| Probe | Read the real codec, resolution, frame rate, duration and audio layout of the source |
| Ladder | Choose which renditions this particular file should produce - before spending a second of CPU |
| Encode | Encode every rendition with live per-rendition progress, on GPU where the host has one |
| Package | Write the adaptive playlists and a machine-readable manifest describing what was made |
| Deliver | Download the whole package as a ZIP, or a single flat MP4 of any rendition |
Forge shows you which renditions a file will produce before the encode starts — so nobody pays for a rendition that was never going to help.
Not a mockup — two frames from a real job running in the Forge workspace.


Seven are possible. Your file gets the ones that make sense for it, and no others.
| Rendition | Who it serves |
|---|---|
| 144p | Fallback for a stalled or metered connection |
| 240p | 2G / edge-of-coverage mobile |
| 360p | The rung most Indian mobile viewers actually settle on |
| 480p | Comfortable mobile and low-end desktop |
| 720p | The default good-network experience |
| 1080p | Desktop, smart TV and connected-TV apps |
| 1440p | Premium source, kept only if the master carries it |
A 720p master is never blown up to 1080p just to pad the ladder. Upscaling costs money and adds nothing a viewer can see.
Renditions are chosen from the short edge, so a vertical reel and a 16:9 OTT episode both come out correctly framed — no letterbox or pillarbox bars baked in.
A master playlist plus one playlist per rendition, aligned so a player switching quality mid-view does not stutter.
Any rendition, rewrapped as one file you can email, edit or upload. Rewrapped — not re-encoded — so the pixels match what the player streams.
A machine-readable description of everything produced, so the systems downstream never have to parse a playlist to learn what happened.
The industry-standard billable-minute model, with the arithmetic shown to you rather than hidden inside an invoice.
100 minutes of output at HD is 200 billable minutes. No per-seat licence, no per-codec option to buy, and self-hosting on your own hardware is on the table.
Rates are quoted per channel — ask us for yours.
Running, and honest about the rest.
| Capability | State |
|---|---|
| Upload, queueing and per-job workspaces | Working |
| Source probe and pre-flight rejection of bad input | Working |
| Adaptive ladder selection, previewed before encode | Working |
| Multi-rendition encode with live progress | Working |
| HLS packaging with correct bandwidth and codecs | Working |
| Flat MP4 of any rendition, by remux - no quality loss | Working |
| ZIP download of the finished package | Working |
| Usage metering in billable minutes | Working |
| Automated QC - loudness, black frame, freeze, VMAF | Planned |
| Captions - speech to WebVTT / SRT, optional burn-in | Planned |
| Push delivery to CDN, S3 or SFTP | Planned |
| Watch folders and a visual workflow builder | Planned |
Transcoding is the first step of the same job, not a separate vendor.
Send one file. We will run it through Forge and send the package back, so you can judge the output rather than the pitch.