Why AI video still needs traditional VFX
The clip looked finished when you first saw it. Then it went into the edit next to real footage and stopped holding up. This explains what is missing, who does that work, and what to ask before you buy AI production.
A generated clip can look finished the first time you see it. Hold it for four seconds instead of one, or cut it against real footage, and it shows flicker, soft edges and detail that drifts.
If you are buying AI production, this is the part to ask about. If your AI vendor cannot tell you who is doing the compositing and the cleanup, the finishing has not been budgeted.
The reason is that a model does not build the thing it shows you. It draws a picture of it and guesses again on every frame. The job of traditional VFX is to make that guess hold still, sit with real footage, and survive a TV delivery spec. I came to AI through post-production: editing, VFX and finishing since 2013, and that is how I look at every generation.
01 - The seam
Material is not a shot
A finished shot has to agree with the frame before it and the frame after it. It also has to agree with a lens, a grain structure, a colour space, an eyeline and a light direction. A generation agrees to none of this by default, because it was never modelling continuity. It was modelling plausibility, one frame at a time.
This is the distinction that decides whether a project ships. On the Gigaclear commercial, the environments coming out of the model were the reliable part. The hard part was the hundred-character crowd, rebuilt in Cinema 4D, and the real actors composited into a world that had no idea they were coming. AI supplied the raw plausibility. Everything that made it a national television spot happened afterwards, in the parts of the pipeline that predate AI by decades.
02 - Time
Temporal consistency and flicker
The clearest way to see the limitation is to slow a clip down and watch a single surface. A brick wall shimmers. A jacket weave crawls. A patch of skin breathes at the pore level. None of this is motion in the scene. It is the model re-deciding what the surface is, frame by frame, with no memory of what it decided a sixteenth of a second earlier. Because the model holds no persistent geometry, it has nothing to be consistent to.
The fix is traditional VFX work. You stabilise, you work in patches, you rebuild a clean plate and re-project it, you deflicker, sometimes you frame-blend or run optical flow to average away the crawl, and where a surface matters to the shot you replace it outright. You are imposing, by hand, a continuity the model could not supply, until the eye stops noticing the boil.
03 - The actor problem
Compositing generated worlds with real footage
The moment a real person enters an AI environment, every weakness in the generation shows, because the actor is real. They obey gravity, hold their proportions, and cast a consistent shadow. The plate behind them does not. So the seam sits in the relationship between the actor and the environment, and building that relationship is the compositor's job.
That relationship is assembled from ordinary, long-established craft:
- Edges and mattes: a hard AI-blur edge reads as a sticker. Real edges have motion blur, spill, and softness that changes with depth. You roto, you refine, you decontaminate the fringe.
- Interaction: contact shadows, bounce light, and reflections that the environment should be casting onto the actor and never does. You paint and light them in.
- Depth: atmosphere, defocus and a shared depth cue so the actor sits in the world rather than in front of a screen of it.
Some shots are better filmed. Close-up product handling and dialogue close-ups are the two I would shoot for real and then build the generated world around.
04 - Anchoring
Tracking, match-move and the plate
Add an element to a moving shot and the question is immediate: does it belong to the camera, or does it float on top of it? AI-generated camera motion is a particular problem here, because there is no real camera behind it. There is no consistent focal length, no true parallax, no rigid relationship between foreground and background. The move looks right and solves for nothing, which means anything you add has nothing trustworthy to lock to.
So you reconstruct what was never there. You track the shot, you solve for a camera where you can, and where the generated motion is too incoherent to solve you stabilise, add your element into the calmed plate, and re-introduce the movement afterward. A logo, a product, a set extension only sits in the plate when it shares the plate's motion. Match-move is how a two-dimensional guess is given a place to stand in three dimensions.
05 - The artefacts
Cleanup, and why the model cannot help
The familiar failures are easy to spot: hands that change between frames, a logo that morphs as the camera passes, or text that resolves into approximate letters. They show the difference between generating a plausible appearance and maintaining a stable object through a shot. A hand has a fixed structure; an image model may produce the appearance of one without preserving every rule from frame to frame.
The repair is old-fashioned paint and tracking. You lock text to a tracked surface or replace it with real type. You rebuild a logo as a clean graphic and track it back on. You patch a warping hand across frames, borrowing from cleaner ones. This is the work that decides deliverability, and it is the work no amount of prompting removes. That is why "95% AI" is an honest description of the material and a misleading description of the hours. The last five per cent is the finishing, and finishing is most of the craft.
06 - Continuity and the ear
Colour, grain, and sound
A single spot is rarely one model. One shot comes from here, another from there, and each arrives with its own idea of black level, contrast and colour temperature. Cut them together untouched and the edit flickers between worlds. So you grade for continuity as well as for look: you neutralise, match, and carry a consistent response through the sequence so the shots agree they were photographed by the same instrument.
Then you make it photographic. Generated frames are often suspiciously clean, or carry the wrong grain, or lack the lens signature the eye expects, the slight softening at the edges, the chromatic fringing, the vignette, the bloom. You add matched grain, a real lens character, and defocus so every shot shares one optical fingerprint. And underneath all of it, sound. A convincing image with dead audio reads as fake; the right footstep, room tone and low-end weight will hold an imperfect frame far longer than another render pass ever could.
07 - The point
Through VFX, not around it
None of this is an argument against AI, which is a fast way to generate useful material. The model’s output then has to join the rest of the production process, and you can see where finishing sits in the seven stages. Generation provides a first answer. Finishing tests whether that answer can hold through an edit and a delivery.
Frequently Asked Questions
Does AI video replace VFX?
No. AI generates material and VFX finishes it. Compositing, tracking, cleanup and grading are what make a generation hold up as a finished shot.
Why does AI footage flicker or fall apart when held longer?
Because a model doesn't render a stable scene. Temporal consistency breaks over time, producing flicker and drift that has to be fixed in post.
How do you combine AI-generated worlds with real actors?
Through compositing and match-move: tracking the plate, anchoring the generated world to it, and matching grain, colour and light so the seam disappears.
What VFX work does an AI shot still need?
Temporal cleanup, artefact removal, compositing, tracking, and continuity of colour, grain and sound. This is the finishing a model can't do itself.
Send three shots. Within a day you get a written reply: what is realistic and how I would make it. This first read is free.
Send the brief → See how the Gigaclear commercial was made →