AI
AI in video production, honestly
What the tools genuinely do to a post schedule, what they cost you if you trust them too far, and how they are used here.
The position
AI is used here for the mechanical parts of post and kept out of the creative ones. Transcription, rough assembly from a transcript, rotoscoping passes, noise removal, upscaling and object removal are all faster and often better than they were done by hand. The structure of a cut, the choice of take, and the shape of a story are not.
That distinction is not a philosophical position. It is what the tools are currently good and bad at, and it will move.
Where it earns its place
- Transcription and text-based rough assembly, which turns a day of logging into an afternoon
- Rotoscoping and matte generation that used to be traced frame by frame
- Speech clean-up: background noise, room tone, and salvaging a take recorded badly
- Upscaling and frame interpolation on archive and low-resolution material
- Object and reflection removal on plates that would otherwise need a full clean-plate build
- Translation and caption drafting for language versions, before a human pass
Where it is kept out
Generated footage is not passed off as shot footage. Where a client wants generative imagery as part of the creative, it is briefed, quoted and labelled as such — not slipped into a cut because it was quicker than a pickup day.
Every AI-assisted pass is reviewed frame by frame. The failure modes are subtle: a roto that breathes, a face that changes between cuts, an upscale that invents detail that was never in the plate. A tool that is right ninety-five percent of the time is still wrong at twenty-four frames a second.
If your brand has a position on AI in production — disclosure requirements, a ban on generative imagery, a rights concern about training data — say so at brief stage and the workflow is built around it.
What it means for a brand's budget
The saving is real but narrower than the noise around it suggests. On a typical corporate or commercial job it comes out of the logging, the clean-up and the versioning, which is meaningful on long-form and on jobs with many deliverables, and close to irrelevant on a thirty-second spot.
Where it changes things most is volume: a brand producing content continuously rather than in campaigns can now run a pipeline that would previously have needed a second editor.
Why this page exists
Software is the other half of this practice — tools built because a job needed something that did not exist yet. That means the AI question here is not answered from a marketing page but from having built with the things. The products are listed on the homepage.
What it costs
R6,500 per day, excluding VAT, wherever the job comes from. What a day covers is set out on the pricing page.
Related
- Video editing Offline and online edit in Premiere Pro, cut to the brief.
- Animation & compositing After Effects compositing, Mocha tracking, clean-up and green screen.
- Motion graphics Titles, lower thirds and animated brand systems.
- Freelance post production The hub: what a day covers, the rate, and how it runs with your team.
Have something to cut?
Send the brief, the deadline and the deliverables. I'll come back with a scope and a quote, or tell you straight if it isn't a job for me.