The Render vs The Result: What a 3D Surgical Preview Doesn't Tell You - A Surgeon's Guide To Simulation

​​The difficulty begins when an illustrative image is quietly promoted—from a shared sketch of intent into a prediction of outcome. That gap, between the render and the result, is the subject of this article. Nothing below argues that the tool should not be used. It argues that its output should be read for exactly what it is.

​

​

First, the Honest Case for the Technology

 

It would be unfair to dismiss surgical simulation as a gimmick, so let me make its strongest case before questioning it.

 

Systems such as Canfield's Vectra XT use multiple synchronized cameras (stereophotogrammetry) to capture a three-dimensional surface; cloud platforms such as Crisalix generate a model from uploaded photographs. For the most measurable procedure—breast augmentation with an implant of known volume and geometry—the published results are encouraging.

 

A 2024 study of 42 patients (84 breasts) using the Vectra XT reported a mean volumetric discrepancy of 21.5 ± 10.3 cc between the simulated and the three-month post-operative breast, which the authors expressed as a mean volumetric accuracy of 91.9%.

 

Patients also tend to value it. In one cohort of 38 women, roughly nine in ten agreed—completely or partly—that their final result resembled the simulation, and the large majority said it helped them choose an implant size. Used well, a simulation can turn an anxious guess about cup sizes into a shared, concrete reference.

 

The evidence is not uniformly glowing, and honesty requires saying so. Some studies find that although patients choose to use 3D imaging, it does not reliably translate into clinically meaningful improvements in patient-reported outcomes.

 

The fairest summary is that a simulation may improve the consultation—communication, confidence, ease of choosing—without guaranteeing a better operation or a closer match between expectation and result.

 

So the useful question is not "Does the tool work?" It is narrower and more revealing: What was measured, in whom, and how closely does that measurement match what the patient believes the image is promising?

​

​

The Headline Number Comes From a Selected Group

 

Read the accuracy studies closely and a pattern emerges.

 

The same 2024 study reporting 92% volumetric accuracy did not measure it on everyone. The researchers excluded patients undergoing additional procedures and, more importantly, excluded patients with tuberous or ptotic breasts—a constricted base or appreciable droop—and stated the reason plainly: "due to limitations of the imaging system."

 

This was disclosed clearly and is not misconduct. Defining a study population so a specific question can be examined reliably is ordinary science. The problem arises only in translation—when a conditional result becomes an unconditional claim on the journey from the journal to the clinic website.

 

The study demonstrated high mean volumetric accuracy in a favorable group. It did not establish that every anatomy, or every visible aspect of the breast, would be predicted that well.

 

And volume is only one component of appearance. Two breasts can carry nearly identical volume while differing visibly in projection, lower-pole shape, nipple position, cleavage, asymmetry, and how tissue drapes over the implant.

 

That is exactly where the same study is more sobering. Alongside the 21.5 cc volume discrepancy, it reported a surface root-mean-square deviation of 4.5 ± 1.1 mm and an anterior–posterior projection difference of 8.82 ± 5.64 mm—a spread almost as large as the average itself.

 

None of this invalidates the technology. It shows that "accuracy" depends entirely on which measurement you pick. On the single dimension a patient studies most in profile—projection—the preview and the outcome parted company more than a headline volume figure suggests. A strong number for volume should never be read as photographic fidelity for shape.

​

 

A Published Disagreement Worth Understanding

 

This skepticism is not confined to outside critics; it played out in print, in the specialty's own journal.

 

An influential 2014 study by Roostaeian and Adams compared simulations with three-month outcomes in 20 patients and reported roughly 90.8% accuracy for volume and, expressed as average surface area, better than 98% for surface contour.

 

A senior and widely respected breast surgeon, Elizabeth Hall-Findlay, then published a formal comment. Her assessment was blunt: in her own experience with the same system, the simulations were "far from 90% accurate."

 

Crucially, she did not claim the software's measurements were wrong—she allowed that "perhaps the numbers measured by the system are accurate"—but she found that they did not capture practical, visual realism. She reported spending consultation time reassuring patients that the image was not an exact prediction and adjusting it toward a realistic result, and she preferred a different tool for setting expectations altogether: actual post-operative photographs of similar patients.

 

Here is the part that a one-sided retelling leaves out. The original authors published a reply. They defended their objective analysis—the data, they wrote, indicated that this conclusion was not supported—and, on the specific claim that 3D imaging lengthened consultations, they reported the opposite in their own practice, attributing the difference to how the system was set up and integrated.

 

Why This Debate Matters: The exchange reveals that the real question is not simply "is the software accurate?" but "which definition of accuracy matters in a consultation?" A model can agree with reality by a volumetric calculation and still produce an image an experienced surgeon considers too round or too lifted. Numerical fidelity and the felt sense that "this looks like what I'll get" are not the same property.

 

 

​

When the Starting Anatomy Changes, So Does the Prediction

 

There is a further finding that deserves careful, not dramatic, handling.

 

A 2017 study compared Crisalix simulations with six-month outcomes in 20 patients, with three attending surgeons and ten residents rating features including height, width, volume, projection, and nipple correction. The simulations were more representative for symmetric breasts than for ptotic or tuberous ones.

 

And the divergence had a direction: for symmetric and tuberous breasts, the real outcome was judged more appealing than the simulation; for ptotic breasts, the real outcome was judged less appealing than the simulation.

 

Sit with the ptotic finding, because it is the one that matters most for the commonest patient. In that subgroup, the image looked better than the eventual result—an over-optimistic preview in precisely the anatomy (some degree of droop, which age and childbearing make near-universal) most likely to walk into a consultation.

 

But it is important not to overstate this. The same study found the opposite bias for symmetric and tuberous breasts, where the simulation was conservative. So this is not evidence that simulations universally, still less deliberately, flatter.

 

It is one small, retrospective, single-platform study based partly on aesthetic ratings. The defensible conclusion is narrower and more honest: baseline anatomy influences how a simulation errs, and ptotic anatomy carries a particular risk of an optimistic preview. That qualification belongs in the consultation room, not just the discussion section—and it belongs there without exaggeration.

​

 

Why Facial Soft Tissue is a Harder Prediction Problem

 

Everything above concerns the breast—the easier organ to simulate. The face is harder, for a reason that is fundamental rather than a matter of better software.

 

Predicting bone is tractable. Bone is rigid; move a planned segment and you know where it goes, which is why virtual planning has become genuinely valuable in orthognathic and craniofacial surgery.

 

Predicting soft tissue is a different class of problem. Skin, fat, muscle, fascia, and scar do not behave as one uniform material; their response depends on thickness, elasticity, direction, underlying support, surgical release, closure, and healing. The models that predict how a face will drape over an altered skeleton are approximations of a material that is uneven, direction-dependent, and biologically active.

 

The numbers reflect this reality. In a 2022 study of 20 patients undergoing bimaxillary surgery, the proportion of the facial surface predicted within ±2 mm ranged from 69.4% to 96.0% across patients, with the lower face least consistent.

 

A 2024 systematic review of 40 studies found whole-face mean errors from 0.27 to 2.9 mm—and, tellingly, the region of greatest error shifted with the operation: around the upper lip and paranasal area after Le Fort I osteotomy; around the lower lip and chin after mandibular surgery; across lips, chin, and paranasal region after bimaxillary surgery.

 

This is not a field that "cannot agree where it fails"—it is a field that has found error to be operation-specific, which is a more precise and more useful truth. Either way, there is no single, stable "facial accuracy" number that can be applied to your face and your procedure.​

​

​

What a Simulation Cannot See: Your Individual Healing

 

There is one variable no standard consultation render models at all, and it is the one that ultimately decides the result: healing.

 

A simulation depicts a finished surface at a single moment. The face or body that actually arrives is the product of a months-long biological process—swelling that resolves unevenly, scar tissue that contracts, cartilage that remembers its old position, skin whose thickness and elasticity shape the final drape, and an inflammatory response that differs in every person and sometimes between the two sides of the same face.

 

Some newer, data-trained systems may embed average patterns of tissue response, but an average is not a prediction of how you, specifically, will heal.

 

A render does not tell you how much you will swell or how quickly it will settle, whether scar will contract unevenly, how cartilage will finally sit, whether a pre-existing asymmetry will remain visible, or whether a complication or revision will change the result. The image is clean because it omits the journey. The real outcome is partly determined by that journey—which is why two patients handed identical simulations can finish in visibly different places.

​

 

The Consent Problem: A Vivid Picture is Not a Caveat

 

Most responsible surgeons say the same true sentence: The simulation is a guide, not a guarantee. The difficulty is that the sentence and the image do not carry equal psychological weight.

 

The caveat is abstract and spoken; the render is concrete, detailed, personal, and seen. It does not look like an illustration of an anonymous patient—it looks like a photograph of you, already operated on.

 

That can create an anchoring effect: once a patient has fixed on a specific endpoint, later discussion is measured against it, even if everyone agreed at the outset that it was only illustrative. We are built to trust a vivid, specific image of ourselves more readily than a cautious sentence spoken over it.

 

This matters ethically because cosmetic surgery is elective and satisfaction depends heavily on expectations—and it may matter legally. Analyses of aesthetic-surgery disputes repeatedly find that the commonest source of conflict is not a technical complication but a gap between the objective result and the patient's subjective expectation—exactly the gap an anchoring image can widen.

 

A 2025 review of Italian aesthetic-surgery jurisprudence described a developing "mixed obligation" model, in which technical competence is weighed alongside prognosis, communication, and the patient's reasonable expectations, with inadequate consent and mismatched expectations recurring in disputes.

 

This should not be generalized automatically to every jurisdiction; legal systems differ. But the underlying risk is not jurisdiction-specific: the more individualized and photorealistic the preview, the harder it becomes to argue that it carried no representational weight.

 

It is worth noting that even the authors of pro-simulation studies flag the need for careful documentation and consent to protect against legal exposure—a quiet acknowledgment that the persuasive image cuts both ways. The remedy is not one more sentence in a consent form. It is communicating the uncertainty as clearly as the proposed improvement.

​

 

The Simulator is Now in Your Pocket

 

Whatever the merits of the clinical systems, the ground has shifted beneath them. Simulation has left the photography suite and moved onto the phone.

 

Free and low-cost apps now generate an "after" image in seconds, and beauty filters perform a real-time version of what an expensive clinical rig does. Patients increasingly arrive having already "simulated" their surgery the night before. The research on these consumer tools is not reassuring.

 

A 2025 analysis of one popular filter documented measurable changes to nasal shape and width, lip height, eyebrow position, mid-face projection, and jaw prominence, alongside skin smoothing and the erasure of tear troughs and folds. The study was small—ten participants—and cannot establish population-wide psychological effects, but it demonstrates that these images are not neutral mirrors: they systematically redraw the face.

 

Unlike older mesh-based filters that simply stretched pixels and often warped background lines, modern generative AI filters use adversarial networks to regenerate the entire image from scratch frame-by-frame. They create flawless, photorealistic alterations with zero visual warping—making their manufactured endpoints far more convincing to the untrained eye.

 

Reviewers of this technology warn that generic algorithms take little account of an individual's skin, ethnicity, or the realities of surgery and anesthetic risk, and may set expectations no operation can meet—and they conclude that such output should be treated as probabilistic and illustrative rather than deterministic. The images themselves carry no such warning. They carry the smooth confidence of a photograph.

​

Related Reading from Dr. Saha:  

A close-range, front-camera selfie measurably enlarges the nose and foreshortens the chin, manufacturing "flaws" that live mainly in the optics. The selfie distorts your starting point; the filter invents your endpoint. Read more in our detailed optical breakdown: Why Does My Nose Look Bigger in Selfies?

 

A patient can now arrive with a complaint the camera invented and a goal the software invented—neither of which corresponds to their actual face, and both rendered more convincingly than anything in the mirror.

​

 

The Fix is a Design Problem, Not Just a Research Problem

 

Here is what makes this, at root, a technology story as much as a medical one.

 

Every other mature predictive system communicates its own uncertainty as a matter of course. A weather forecast shows a cone of possible paths, not a single line. A financial projection shows a fan of scenarios. A clinical risk calculator reports a range or a probability. Displaying uncertainty is a solved problem in information design.

 

A surgical simulation, by contrast, usually shows one crisp, deterministic image. That is not an unavoidable technical limitation. The systems' error characteristics are documented—the accuracy studies exist, the failure modes are named, the anatomies where prediction is weakest are known.

 

A more candid interface could show several plausible outcomes instead of one; a range for projection or volume; blurred or flagged regions where soft-tissue prediction is least reliable; optimistic, central, and conservative views side by side; a warning when a patient's anatomy falls outside the population the system was validated on; and a permanent, visible label that the image is illustrative.

 

Most interfaces do none of this. That most do not is not proof of intent to deceive—but it does mean the design can make precision far more visible than uncertainty, and it does so in a commercial setting where the same image helps decide whether an elective operation goes ahead.

 

In that context, suppressing visible uncertainty is not a neutral choice. The next advance in surgical simulation should not be photorealism alone. It should be honest uncertainty.

​

 

How to Look at a Simulation You Are Shown

 

A simulation can genuinely help you—if you read it correctly. Adopt these practical habits during your consultation:

 

1.  Treat it as a conversation, not a contract. Its job is to help you and your surgeon describe the same goal in the same language. It is not a preview of your guaranteed result.

2.  Ask the decisive question: "Can you show me real before-and-after photographs of past patients whose starting point looked like mine?" Real outcomes in comparable anatomy are worth more than any render—and a willingness to show them, rather than only the clinic's most dramatic cases, is a good sign.

3.  Ask where the simulation is least reliable for your anatomy. A surgeon who can answer specifically understands the tool's limits. Vague reassurance is itself information.

4.  Ask what the software predicts best. Volume, surface contour, and overall profile appearance are different outcomes with different clinical reliability.

5.  Distrust flawless symmetry and perfect skin in a render. Real bodies are not symmetrical and real skin has texture; a too-perfect image is a warning, not a target.

6.  Never accept a filtered or app-generated image as a surgical specification. A consumer filter redraws several structures at once, with no regard for what an operation can safely reproduce.

7.  Judge the eventual result the way it was planned—in a mirror and in properly taken, standard-distance photographs—not in a close-range selfie, which will distort a good outcome for the same optical reasons it distorted your original concern.

 

​

When to Be Most Cautious

 

Be especially wary of the persuasive power of a render if:

 

-   The simulated image, rather than a physical concern you feel in the mirror, is the main thing driving your wish for surgery.

-   You find yourself returning to the image repeatedly, or obsessively measuring your real face against it.

-   Your reference goal is a filtered or AI-generated version of yourself rather than an accurate photograph.

-   You feel significant distress about a feature that looks unremarkable in an accurate, well-lit photograph taken from a normal distance.

 

These overlap with the warning signs of Body Dysmorphic Disorder (BDD)—a recognized and treatable condition in which distress attaches to a perceived flaw out of proportion to what others see. It is not rare in this setting: meta-analyses estimate its prevalence among people seeking aesthetic procedures at roughly 15–25%, far above the general population.

 

Where that pattern is present, the responsible first step is a conversation with a mental-health professional, not an operation—a point now embedded in formal cosmetic-surgery clinical guidance across several countries.

​

 

In My Own Practice

 

I use imaging in planning and I find it useful—so I hold a declared interest in this technology, and you should weigh what I have written with that in mind.

 

But I do not let a simulation carry the weight of a promise, and I do not plan any procedure on the strength of a render alone. When I show a patient an image, I say plainly what it is: a shared sketch of an intended direction, not a photograph of a fixed destination.

 

Wherever possible, I show real results from past patients with a comparable starting point, because those are the honest evidence of what surgery in that situation actually achieves—including its limits. And I try to spend more time on what surgery cannot do than on the seductive picture of what it might.

 

What most people want, in my experience, is not the flawless face on a screen. It is to stop being bothered by one feature every time a camera appears. That goal is reached by respecting the architecture of a real face and a real body—not by chasing a rendered ideal that healing was never going to deliver.