Stitch and assemble
Join ordered clips or trusted render shards with ffmpeg.stitch_export.v1 and ffmpeg.assemble.v1.
Apple Silicon media compute
An Apple Silicon render farm for real media work.
RenderMac turns available Apple Silicon capacity into an observable API for video stitching, caption burn-in, transcode, audio mix, and Remotion rendering. Buyers submit a named service, receive a capacity estimate, and retrieve a verified artifact.
Current availability: RenderMac's dogfood release is private-first-party and organization-private first. Public provider capacity stays disabled until isolation, payments, and physical-fleet gates pass.
What the render farm does
Encode on Apple Silicon
Use a constrained transcode contract with VideoToolbox-aware admission instead of sending arbitrary shell commands to a Mac.
Burn captions and mix audio
Apply declarative ASS/SRT caption inputs and bounded timeline audio-mix settings with verification before settlement.
Render Remotion timelines
Bind composition metadata first, then split an authoritative frame range into provider-executed shards and assemble the result.
Why named services matter
A generic “remote Mac” is difficult to price, secure, and retry. RenderMac keeps the buyer contract explicit: each service has a strict input schema, a runner identity, capacity requirements, a quote, a bounded lease, and output verification. That makes a flaky distributed provider pool feel closer to a normal media API.
- Quote before funding, with ETA and fee bands.
- Retry a timed-out create with the same idempotency key.
- Receive phase events or signed terminal webhooks.
- Download through short-lived artifact URLs; media does not proxy through the control plane.
- Keep sensitive workloads in an organization-private capacity cell.
Apple Silicon render farm vs. a generic cloud VM
RenderMac is designed for workloads that benefit from Apple-native codecs, Metal-adjacent tooling, or a Mac-based rendering environment. It is not positioned as a universal replacement for every CPU, GPU, or long-running batch workload. The service catalog and capacity quote are the source of truth for what can run.