New to post? Start here.

How it actually works

A visual-effects shot isn't a file you nudge. It's a build, assembled against one specific range of frames. The camera move is solved across those frames. The roto — the hand-drawn outlines that separate an actor from what's behind them — is drawn across those frames. The elements are timed to land on those frames.

Move the cut and the shot no longer starts where the build starts. Extend it by twelve frames and someone has to solve, roto and comp twelve frames that don't exist yet, then re-render, re-check and re-deliver the whole shot to see them. Trim it and the work is still spent, just thrown away. This is also why studios ask for handles — a few extra frames either side of each cut, usually 8 to 12 — so small moves can be absorbed without reopening anything.

What it costs when it slips

The expensive part is rarely the extra frames. It's that reopening a finished shot restarts everything around it: another render, another internal check, another round of client review, another delivery, another conform. A change worth a few frames of work can cost a day of everybody's attention — and it arrives at the point in the schedule where there is least of it left.

What to do about it

1. Lock the VFX sequences first, not the whole show. Partial lock is real and useful. If the dialogue scenes are still moving but the big sequence is settled, say so — that sequence can start.
2. Send a change list, not a new cut. "Shot 40 is eight frames longer at the head" is actionable in minutes. A fresh export with no notes means someone has to find the differences before anyone can act on them.
3. Batch changes into one round. Five notes delivered together cost one cycle. The same five notes trickled over a fortnight cost five.
4. If a sequence genuinely can't lock, say so at bid time. Fluid work can be priced and planned as fluid. What can't be absorbed is a fixed price that quietly turns out to be a moving target.

What makes one frame harder than another

Producers reasonably ask "what's a shot cost?" — but shot count is the weakest of the three numbers. The same second of screen time can be an hour or a fortnight depending on four things:

Does the camera move? A locked-off camera means the background can be built once. A moving camera means the scene has to be solved in three dimensions before anything can be placed in it.

Does something cross in front? An actor, a branch, smoke, a hand. Anything that overlaps the thing being replaced has to be separated by hand, frame by frame. This one line item quietly outweighs most others.

Is there anything soft? Hair, water, glass, fire, motion blur, fine fabric. Hard edges are cheap. Soft edges are where the days go.

How hard does it have to hold? A dark two-second cutaway forgives a lot. A bright, wide, eight-second establishing shot forgives nothing, and that difference is real money.

The multiplier nobody budgets for

Count and complexity are estimated at bid time. Iterations are not — and they are usually what moves the final number. A shot planned for two rounds of feedback that takes six hasn't tripled in difficulty; it has tripled in attention: six renders, six reviews, six sets of notes, six deliveries. Most overruns are not a shot that was harder than expected. They are a shot that was decided later than expected.

What to do about it

1. Ask for the bid split by complexity band, not a flat average. A per-shot average hides which shots carry the risk — and those are the ones worth a conversation.
2. Ask what the number assumes. Every bid rests on assumptions about what will be shot and how. Those assumptions are the actual price. Read them before the total.
3. Settle the reference before work starts. An agreed image the director has actually signed off removes whole rounds of "not quite" — the most expensive note in post.
4. Ask which shots you don't need. The cheapest visual effect is the one solved on the day with a different lens, a different angle, or a real object.

Why the shoot day decides the post cost

A camera records a flat image. Almost everything post does — putting a building behind an actor, removing a rig, extending a set — requires rebuilding the three-dimensional truth that the flat image came from: where the camera was, how it moved, what lens it saw through, where the light fell.

Every piece of data captured on the day is a piece of that truth handed over for free. Every piece not captured has to be inferred, by a person, frame by frame. That is the whole economics of it. A shot that arrives with its data is craft. A shot that arrives without it is archaeology.

The four that cost the most

The actor stood too close to the greenscreen. Green light bounces onto their shoulders and hair, and the automatic separation stops working — so it gets drawn by hand instead, on every frame.

Nobody shot the clean plate. Ninety seconds of the empty set after the take turns a removal into a copy-and-paste. Without it, the background behind the removed object has to be painted, on every frame.

The camera report is blank. Lens, height, tilt and distance take thirty seconds to write down and can otherwise take a day to reverse-engineer.

Nobody called. The improvised "we'll fix it in post" shot is the single most reliable way to spend money you hadn't planned to spend.

What to do about it

1. Put one person in charge of VFX data. Not a department — a person, with the checklist below and permission to hold the camera for thirty seconds.
2. Treat the clean plate as part of the take. Not something to get to if there's time. It is the cheapest insurance on the call sheet.
3. Photograph the light, not just the set. Wide stills, and a grey and chrome ball if anything computer-generated has to sit in that frame. Anything built later has to match a light that no longer exists.
4. Ring your supervisor from set before improvising. Two minutes on the phone usually ends with a cheaper way to shoot the same idea.

What the supervisor is actually doing while you talk

They are sorting. Every idea you describe lands in one of three piles: not a visual-effects problem at all — solvable with a lens, an angle, a real object or a different location; a cheap one — real work, but predictable; and an expensive one — the handful of shots that will carry most of the budget and nearly all of the risk.

Then they are looking for the thing behind the expensive pile. It is usually one creative intention — a particular camera move, a particular time of day, a particular reveal — that is quietly driving five or six shots at once. Find it early and there is often a version of the same idea that costs a fraction as much and reads identically on screen. Find it in the grade suite and there isn't.

What to bring — and what it's for

1. The script, or the cut. Whichever exists. A rough assembly is worth more than a polished treatment, because timing changes what a shot costs.
2. The references you love. Not for flattery — a picture settles in ten seconds what a paragraph argues about for a week.
3. The delivery date, honestly. Schedule is a cost input, not an administrative detail. The same work compressed costs more, and it is better to know that now.
4. A budget band you can say out loud. Not a final figure. A range lets a supervisor design an approach that fits it, rather than quoting an approach you were never going to buy.
5. Your known constraints. The location you've lost, the actor you get for two days, the stunt the insurer won't allow. Constraints are where the good cheap ideas come from.

What you should get back

A first conversation should leave you with four things: which of your shots aren't VFX problems, which few carry the real cost, what to shoot differently to make them cheaper, and where the risk sits if the schedule moves. If you leave with only a number, you haven't had the meeting yet — you've had a quote.