Pitch a Visual Effect Before Its Method Is Solved
Pitch a Visual Effect Before Its Method Is Solved
There is a moment in treatment writing when a technical word becomes the most attractive thing on the page. The image is clear, the room is nodding, somebody asks how it happens, and "we'd do that in CG" arrives like a handrail.
It isn't one. It's a guess in a confident outfit, and what it does is close a question before anyone has asked it properly. The cost shows up later, when a collaborator has spent a week building toward a thing nobody actually described.
The treatment doesn't have to solve the effect. It has to describe the viewing experience precisely enough that a decision can be made, and it has to leave the production unknown standing in the open where everyone can see it. That's a smaller job than it sounds and a harder one than it looks. Three things go in: the ordinary state of the world before anything changes, the change as it would appear on screen, and what a character or an audience can do or understand afterward that they couldn't before. One thing follows from those: the set of relationships the illusion depends on and cannot break. One thing is set aside, clearly labeled: the method.
The example I'll work through is invented for this piece — a conceit, not a proposal for a real package, and nothing in it has been built or tested. Treat it as a paper exercise in how to phrase a pitch.
Write the effect as an event in the scene
Three sentences, in this order, and you'll know whether you have a shot or a mood board. What is present before the change? What changes, visibly? What can somebody do afterward?
"Surreal transformation" fails all three, because it names a category rather than an event. Nobody can approve a category, cut it, or argue with it. A sentence like the parcel's printed stripe lifts off the paper and becomes a handle can be argued with immediately, which is exactly what you want.
Here is the invented version. A parcel sits on a counter, wrapped in kraft paper printed with a single straight red stripe running across its top face, flush with the paper like a design that came from the printer. Nothing else about it is unusual. Then the stripe lifts off the surface along its length — the paper beneath it intact, no tear, no lifted fiber — and curls up into a closed loop standing above the parcel. The two ends of the stripe stay attached to the paper where they always were. A courier slides two fingers through the loop, lifts, and the parcel comes off the counter and swings as he walks out of frame.
That's the before, the change, and the after, and notice how little of it required knowing anything about production.
Now the harder sentence: why is this here? A spectacular image with no consequence is a portfolio piece that wandered into a script. So state what the shot does for the argument of the spot in one sentence, without using the word cool. For the parcel, the sentence might be: the object acquires a handle at the moment carrying it matters. Now the transformation is doing work. The ad's claim happens on screen, in the courier's hand, instead of in the voiceover on top of it. If the only sentence you can write is because it looks incredible, you're pitching an image, not a shot, and the room will feel the difference even if nobody says so.
There's a useful side effect of writing the event first: the ordinary state does a lot of the persuading. Stating that the stripe is printed, flat, and part of the paper's design tells the viewer what the baseline is, and a baseline is what makes a change legible. If you skip it — if the first frame the audience sees is already mid-transformation — you've thrown away the setup that gives the effect its size.
Identify what must remain continuous
An illusion is not an object. It's a set of relationships that have to survive every frame, and your job in the treatment is to name them, then sort them into two piles: what the illusion needs, and what you'd merely like.
For the parcel, the essential list is short.
Contact. The loop has to read as something fingers can pass through, and the grip has to be on the loop rather than on the air near it. If the hand's relationship to the loop is wrong, the audience doesn't see a worse version of the same event — they see a different event, a hand holding nothing while a shape hovers nearby.
Attachment. The two ends of the stripe stay joined to the paper. No gap opens, no seam appears, the loop never floats free of the parcel. This is the relationship that decides whether the audience believes the parcel is being carried or merely accompanied.
Mass. When the parcel lifts, its motion has to match the loop's apparent stiffness. A loop that flexes like wet string attached to a box that swings like a box is two objects pretending badly to be one. You don't have to solve the physics; you do have to decide which reading you want and say so, because the two are not compatible and somebody down the line will pick one by accident if you don't.
Plane. A loop reads as a loop only from an angle where you can see through it. Turn the camera until the loop's plane is edge-on and it collapses back into a line. That's the same stripe, the same parcel, a different event.
Timing. The hand arrives as the loop reaches the top of its rise. That relationship holds whether or not the loop exists on the day of the shoot, and it's a dependency on the performance as much as on anything else.
Then the second pile, which is where most treatments accidentally spend their words: the exact red, the height the loop rises to, the direction it tilts as the hand approaches, the speed of the curl. These are choices. Change any of them and you get a different-looking version of the same event. Useful to state, but they're not what the effect rests on.
The test for the sorting is one question per item. If this were violated, would the audience see a different event, or a worse-looking version of the same event? A floating loop is a different event. A teal loop is the same event in the wrong palette. Mark which pile each line belongs to and your collaborators can tell, in about nine seconds, which constraints are load-bearing and which are taste.
That's the whole gift you're giving them: a precise problem that survives being handed to somebody else. What you are not giving them is a rig, a simulation approach, or a recommendation to build it practically. Those are answers, and you don't have one yet.
Show the limit of each planning image
The most persuasive object in a treatment is also the most dishonest one, and it's the finished concept frame.
Picture the concept still for the parcel: the loop standing above the counter, scaled to weight-bearing proportions, the courier's fingers already through it, warm light across the kraft paper. It is probably beautiful. It shows the destination, and it demonstrates nothing whatsoever about the stripe leaving the paper — which is the hard part. If the difficult moment is the transformation, then a still of the finished handle has answered the easy half of the question while looking like it answered the hard half. The audience of a static image does the in-between work for free, in their own heads, and arrives convinced by effort they supplied themselves.
So build states instead of a beauty shot. Four frames, roughly drawn, deliberately unglamorous if that's what it takes to keep them honest.
State one: rest. The stripe flat, flush, printed. The parcel ordinary.
State two: rise. The stripe peeling along its length, the two ends still touching the paper. This frame can be an arrow and a shape. It does not need to be beautiful. It needs to show where the connection stays.
State three: close. The loop formed, the hand in it, the ends still attached.
State four: lift. Parcel off the counter, hanging, swinging.
Then annotate each frame with what the audience is supposed to read, not what the equipment is supposed to do: ends stay attached; loop closes into a shape a hand can enter; weight reads in the parcel's motion.
And label the material accurately. An isolated still is a statement about appearance. It is not a demonstration of motion, and it is not a demonstration of integration with a plate. A rough animatic of the four states is a better tool — it can communicate the timing and the readability of the change — but even a moving rough shows an intention about motion. It does not show that any production route achieves it. Two different documents, two different claims, and the moment you blur them you've made the unresolved thing look resolved.
Failure has a signature, and it's short: the word seamless beside an endpoint image. Seamless is a feeling, not an instruction. Nobody can check it. It goes in the treatment, it sounds like rigor, and it quietly tells a producer that the difficult part has been handled.
Turn the unknown into a decision-relevant question
"How do we do it?" is not yet a question. It doesn't tell anyone what changes depending on the answer, so it produces a discussion rather than a decision.
For the parcel, the genuine unknown is how the stripe leaves the paper and how the ends carry load. Two broad families of answer exist, and this is not the place to choose between them — one where the loop is a real object that the camera and the edit hide, one where the parcel, the stripe, and the lift are all generated, and honestly also the space in between. Choosing is what the comparison piece is for. Your job here is to isolate the part of the unknown that changes the film if the answer goes one way rather than the other.
Here is a question shaped for that: does the written camera move keep the loop's plane and both attachment points readable for the whole lift, or does it swing the loop edge-on partway through?
Follow it through. If the answer is yes — the written move keeps the loop's plane and both ends readable — the lift can be staged as written, but the camera cannot arc during it: the courier's lift has to happen roughly parallel to the lens. That is Version B below. If the answer is no — the move swings the loop edge-on partway through, say the camera arcs around the courier during the lift — then the loop turns toward edge-on and reads as a plain line again, and the event changes rather than just the shot's look. That is Version A, the lens the treatment wanted, and the cost is not a production adjustment invisible to the audience. It's a change to what happens on screen.
So draw the conditional version. Version A is the lens the treatment wanted: it can arc, but partway through the loop turns edge-on and reads as a plain line. Version B keeps the loop's plane and the two attached ends readable for the full lift — which means the camera can't arc during it, and the courier's lift has to happen roughly parallel to the lens. Most of the time, B is quieter than A. Sometimes it's better. Either way, the loss is yours to own, not the effects collaborator's: if the version that can be read is also the version that's boring, that's a creative problem in the idea, and no vendor conversation will fix it.
You can also ask for something small and cheap that answers one question instead of all of them. A still test, using an actual loop of paper or wire taped to an actual box, photographed from the two extreme angles of the proposed move, would tell you whether the loop and the two attachment points still read from where the camera would be. It answers readability. It says nothing about whether a stripe can leave a paper surface, and it shouldn't be presented as if it does. That's the point of a bounded test: it retires one unknown and leaves the others standing.
And if the lift turns out to be unreadable from every angle available to you, you don't have an unresolved method. You have an unresolved shot. That's a different conversation, and it's worth saying out loud before anybody quotes a schedule.
What the treatment says in the end
The desired event, plainly: the stripe lifts, the ends hold, the hand goes through the loop, and the parcel swings like a parcel. The continuity the illusion stands on: contact, attachment, mass, plane, timing. The unresolved thing, phrased so that somebody can answer it and so that its answer costs something specific.
Note what's missing. No method word. No seamless. No claim that the shot is solved, because it isn't, and a treatment that says otherwise will be believed at exactly the wrong moment.
Everything else — how, on what schedule, at what price — belongs to a conversation that starts with a question worth answering. What the treatment owes the reader is an image exact enough to want, and honest enough to investigate.
Frequently asked questions
What belongs in a treatment for a visual effect before its method is solved?
The body says three things go in: the ordinary state before anything changes, the change as it would appear on screen, and what a character or audience can do or understand afterward. From those follows the set of relationships the illusion depends on and cannot break. The method is set aside, clearly labeled.
How can a writer tell a load-bearing relationship from a preference?
Ask whether violating it would make the audience see a different event or merely a worse-looking version of the same event. In the parcel example, contact, attachment, mass, plane, and timing are essential; exact red, height, tilt, and speed are choices. A floating loop is a different event; a teal loop is the same event in the wrong palette.
Why is a finished concept frame a poor way to pitch the hard part?
It shows the destination, and a static image lets the audience do the in-between work in their own heads. For the parcel, a still of the finished handle answers the easy half while looking like it answered the transformation. Build states instead—rest, rise, close, lift—annotated with what the audience should read, not what equipment should do.
What makes a production unknown decision-relevant?
It must be phrased so its answer changes something specific. For the parcel, the question is whether the written camera move keeps the loop’s plane and both attachment points readable for the whole lift, or swings the loop edge-on partway through. Yes means the lift can be staged as written but the camera cannot arc; no means the camera arcs and the loop reads as a plain line again.
What can a small physical test answer, and what can’t it answer?
A still test using an actual loop of paper or wire taped to an actual box, photographed from two extreme angles, could tell whether the loop and two attachment points read from where the camera would be. It answers readability only; it says nothing about whether a stripe can leave a paper surface and should not be presented as if it does. If the lift is unreadable from every available angle, the body says that is an unresolved shot, not an unresolved method.