What Latex Knows That Pixels Don't: Practical Effects vs CGI

Watch Rob Bottin’s creature work in The Thing (1982) beside almost any digital monster of the last decade and the older film wins on a metric that has nothing to do with nostalgia. The dog-thing splitting open, the defibrillator scene, the head that sprouts legs and scuttles away: these were built out of latex, mechanics, gelatine and enough karo syrup to drown a small town, and they still make audiences flinch forty years on. A great deal of expensive digital horror from the 2010s already looks dated. The gap is not a matter of taste or of one era being braver. It is physics, and understanding the physics tells you when to reach for the rubber and when the render is genuinely the better tool.

Light does not lie about a real object

The single largest advantage practical effects hold is that they exist in the same space as everything else in the frame, under the same lights, casting the same shadows. When a lamp on set illuminates a Bottin creature, the light behaves correctly because it has no choice; it bounces off wet latex, pools in the recesses, scatters through translucent silicone exactly as it would off real flesh, because at the level of a single photon the fake skin and the real skin are obeying the same rules. The camera records all of that for free.

A digital creature has to have every one of those interactions calculated and faked. Subsurface scattering, the way light penetrates skin and glows back out, is one of the hardest things in computer graphics to get right, and for years it was simply approximated, which is why so much early CGI flesh looks like painted plastic. The render has to decide how light behaves. The rubber puppet already knows. That is why Chris Walas’s transformation work in The Fly still turns stomachs: Brundlefly is a physical, oozing, fingernail-shedding object that the camera photographed under real light, and no amount of resolution has let a render farm reliably match the way a genuine wet surface throws back a highlight.

The same logic runs through the whole video-era canon of body horror. Screaming Mad George’s climax for Society works because the “shunting” is a real, glistening tangle of bodies that the actors are physically embedded in, and Shinya Tsukamoto’s Tetsuo welds actual scrap metal to actual skin under harsh monochrome light. The horror is legible because the eye trusts what the light is doing.

The actor was really looking at it

The second advantage is on the performance side, and it may matter more than the light. When an actor shares a set with a physical creature, an animatronic head or a puppeteer’s rig, they have something concrete to react to. Their eye-line is correct because there is a real object where the eyes are pointing. Their fear registers as fear of a thing in the room. Steven Spielberg understood this precisely on Jurassic Park (1993), which is remembered as a CGI landmark but is mostly Stan Winston’s animatronics; the T. rex that attacks the cars in the rain is a full-scale hydraulic puppet, and the reason that sequence has aged like granite while later digital-dinosaur films look weightless is that the actors were genuinely being rained on next to a snarling twenty-foot machine.

The Alien franchise is the whole argument in one series. Ridley Scott’s original monster in 1979 was a man, Bolaji Badejo, inside H. R. Giger’s biomechanical suit, photographed in near-darkness so the eye completes what it cannot quite see, and the chestburster that erupts from John Hurt was a real puppet rigged with a compressed-air pump and a bucket of offal, sprung on a cast who had been kept partly in the dark about the mess. The horror of both is the horror of presence. By Alien: Resurrection (1997) the series was reaching for a fully digital “newborn” hybrid, and for all the money on screen it lands soft, because the thing has no weight and the actors are plainly performing at empty air. Same universe, same designers’ DNA, opposite result, and the variable that changed was whether the monster was in the room.

Compare the effect of an actor performing against a tennis ball on a stick, to be replaced in post by a creature they never saw. Even a gifted performer is guessing at scale, weight and distance, and the guess shows in the eyes. The most convincing digital creatures of recent years, the ones built through performance capture, work precisely because they smuggle a real actor’s presence back into the pixels; there is a human being making real decisions underneath. The failures are the fully synthetic monsters an actor had to imagine.

There is a cautionary tale attached to The Thing itself. The 2011 prequel was shot with an extensive suite of practical creatures built by a specialist shop, and then, late in post-production, most of that work was painted over with CGI. The digital replacements are widely regarded as the weakest thing in the film, and enthusiasts have circulated the original practical designs precisely to mourn what was buried. The same studio nervousness that softens horror endings, which I wrote about in why every horror remake softens the ending, reaches into the effects bay too, and it tends to reach for the tool that can be changed at the last minute.

Weight, and the tell of perfect motion

The third difference is weight, and it is the one audiences feel without being able to name. A physical object has mass. When a puppet lunges, it has to overcome its own inertia; when it settles, it settles with the slight overshoot and jiggle of a real thing under gravity. Early animation software had no idea about any of this, so digital creatures moved with an uncanny frictionlessness, accelerating and stopping as if they weighed nothing, because in the machine they weighed nothing. The human eye is exquisitely tuned to biological motion, having spent a few million years learning to read the gait of predators, and it flags weightlessness as wrong long before the conscious mind works out why.

There is a famous accident that proves the point in reverse. Spielberg’s mechanical shark on Jaws (1975) barely worked, so he was forced to keep it off screen and imply it through a fin, a barrel and John Williams’s two notes, and the film became more frightening for the constraint. A creature that misbehaves on set at least misbehaves with real mass; a digital one that misbehaves simply gets re-rendered until it is smooth, and smoothness is the enemy of dread. This is also why the appropriate uses of CGI tend to be things that never had convincing weight to begin with. The liquid-metal T-1000 in Terminator 2 (1991) remains one of the most persuasive digital effects ever made because it is meant to be an impossible, physics-defying substance; the medium’s weightlessness is a feature. Water, fire, morphing, vast impossible scale: these play to the render’s strengths. A creature that is supposed to have muscle, breath and heft plays to its weaknesses, which is exactly where so much horror lives.

The lesson the great practical artists internalised, from Dick Smith through Rick Baker to Bottin and Winston, is that horror is a close-up medium. It wants texture, sweat, the pore-level imperfection that reads as alive. Baker’s transformation in An American Werewolf in London (1981) still holds up because it is shot in bright, unforgiving light on real prosthetics that bulge and stretch, daring you to find the seam. A render can add infinite detail, but it has to invent every flaw deliberately, and invented imperfection has a way of looking designed rather than grown.

The right tool, and why the argument keeps returning

None of this is an argument that CGI is worthless. The false binary is the real enemy. The finest effects work has always been hybrid: Winston’s animatronics finished with a few digital extensions, a practical creature composited into an impossible environment, a physical puppet given a digital eye-blink. The question a good effects supervisor asks is which tool the shot needs rather than which camp to belong to, and horror, with its reliance on close scrutiny and physical dread, skews harder toward the practical than almost any other genre. I made a related case about creature design specifically in the death of the man in the suit, and the two arguments rhyme: the more the audience is invited to look, the more the fakery has to survive looking.

The reason the debate never dies is that the economics keep pushing the wrong way. A physical creature is expensive, slow, and locked in once it is shot; a digital one is a file that can be revised until the deadline, which is precisely why nervous productions prefer it and precisely why so much of it looks provisional. Latex commits. That commitment is why the old monsters still breathe and why every few years a new film, tired of weightless digital dread, drags the foam latex and the karo syrup back out of storage and reminds everyone that the most frightening thing you can put in front of a camera is a real object the actor is genuinely afraid to touch.