You are comparing Fit3D and ShapeScale for a clinic or a fitness business. Both are optical 3D. They do not capture the same way, and they do not estimate composition the same way. This page is for a gym or studio, a med spa, a plastics practice, a wellness clinic, or a GLP-1 program. It is not for the member looking up a $30 gym scan.

Who should buy which

• Keep a Fit3D ProScanner if you already run one (common on gym floors), you sell posture or body-shape scores, and you will only use the fat number as a trend under a locked stance.
• Buy ShapeScale if the consult or trainer walkthrough has to show a photoreal body plus girths, volumes, composition trained on iDXA, and a PDF they can take.
• Do not treat SNAP, Fit3D's $100/mo tablet, as the ProScanner papers below. We did not find a SNAP-vs-DXA study.

What they see in the room

On Fit3D the person being scanned turns on a plate. You get a mannequin and a dashboard on a laptop. The fat % is estimated from the inches, not from tissue. Same class as Styku.

On ShapeScale they stand still. You get a photoreal mesh, girths, volumes, and heatmaps on iPhone or iPad during the consult or floor walkthrough. Staff can also export a PDF of one scan or a two-scan compare and share it from the app. Composition is predicted from that full 3D mesh by a deep-learning model trained on paired scans vs GE Lunar iDXA, not from an inch equation.

What the room has to hold

On a Fit3D ProScanner the person turns. The hardware is a camera tower plus a 38-inch-wide turntable. Fit3D's own help page asks for a clear floor of 78 by 62 inches (about 6.5 by 5.2 ft), plus a 1-foot keep-clear around the unit, no sunlight on the cameras, and no mirrors in view. That does not include the laptop and cables the install actually uses.

On ShapeScale the person stands still on a 2-foot base. The sensor head moves. The room you need is that orbit: 5 ft of rotation clearance, 6 by 6 ft recommended, 8 ft of ceiling. One outlet. No second piece of furniture for a dashboard PC.

Standing still is not a comfort line. A turning plate asks the person to balance while the surface is measured. Independent turntable work (not a Fit3D unit) has measured local body motion up to about 2 cm in one revolution. The sensor-moves capture does not need that balance act, and it does not need a tower parked a fixed distance from a plate.

Fit3D vs ShapeScale

Method

• Fit3D ProScanner: Rotating platform, fixed tower, handles. Composition from anthropometrics: a deterministic equation from inches and surface volume (Fit3D FAQ).
• Fit3D SNAP: Android tablet, no turntable. Same reporting backend (vendor). No SNAP-vs-DXA paper found.
• ShapeScale: Sensor head moves around a still person. Deep-learning composition from the full 3D mesh, trained against GE Lunar iDXA (current model; March 2026 white paper).

What you show in the room

• Fit3D: Mannequin plus a posture report on a laptop dashboard. Results live in a web dashboard. Posture is often ready a day after the scan.
• ShapeScale: Photoreal 3D, girths, volumes, and heatmaps on iPhone or iPad during the consult, plus a PDF they can take (one scan or a two-scan compare). On a new scan you can move the girth location.

Measurements

• Both: regional circumferences and regional volumes. ShapeScale lists neck, shoulders, chest, biceps, waist, hips, thighs, and calves, plus volumes for arms (one combined arm volume in Compare, Body, Heatmaps, and the PDF), torso, and legs.
• Fit3D only: posture analysis (tilt, shift, and ProScanner balance). ShapeScale has no posture score.
• ShapeScale only: photoreal mesh and volume heatmaps on iPhone or iPad. Composition extras that ship today: fat mass in lb or kg, lean mass index, appendicular lean mass, visceral fat as a mass, Health Score, and Shape Score.

Body fat vs DXA

• Fit3D: Not in that white paper. Independent ProScanner papers: Ng n=39 root-mean-square error (RMSE) 5.3 lb (2.4 kg) fat mass; Tinsley (Fit3D + Styku + two others) BF% RMSE 3.7-6.1 pp vs a 4-compartment model (4C); Wong %fat change R2 0.23-0.25 vs Hologic.
• ShapeScale: mean absolute error (MAE) 1.89 pp, R2 0.91 (white paper)

Lean / appendicular lean mass (ALM)

• Fit3D: Vendor estimate from surface measurements. Ng fat-free mass (FFM) RMSE 4.9 lb (2.2 kg) on n=39 researcher equations.
• ShapeScale: lean-mass MAE 3.3 lb (1.49 kg); appendicular lean mass (ALM) MAE 2.1 lb (0.97 kg) (white paper abstract)

Fit3D does not report appendicular lean mass. A whole-body lean or fat-free number is everything that is not fat: muscle, organs, water, and gut contents. It can move when someone drinks or eats. Appendicular lean mass (ALM) is lean tissue in the arms and legs, where most skeletal muscle sits. That is the number a GLP-1 or training program actually wants: did the limbs keep the muscle. ShapeScale reports ALM today. The March 2026 white paper tested it against GE Lunar iDXA (MAE 2.1 lb / 0.97 kg, R2 0.96).

Best use

• Fit3D: Posture programs and gym-floor visual. The fat % is estimated from the inches, not a DXA substitute.
• ShapeScale: Circumference, volume, and composition proof in gyms, studios, plastics, med spa, GLP-1, nutrition, plus a take-home PDF. No posture score.

What Fit3D actually is

The hardware clinics and gyms mean by "Fit3D" is the ProScanner: a rotating platform in front of a fixed camera tower. The patient steps onto the turntable, holds the handles, and stands still while the plate turns. Wong et al. timed one revolution at about 45 seconds on a ProScanner v4.x. Fit3D's own buyer guide says more than 450 data points in under 35 seconds.

The output is circumference, volume, posture, a Body Shape score, and a mannequin-style avatar. Neither Fit3D nor ShapeScale is a DXA scan. Fit3D's own FAQ says the company "algorithmically correlate[s] anthropometric measurements" from the scan to DXA readings, and that the algorithm is "primarily volumetric/anthropometric based." That is a deterministic equation: inches and surface volume in, a fat number out.

Same class as Styku: the patient turns on a plate in front of a fixed camera. The fat % is estimated from those inches. The avatar is a mannequin. The install is a tower, a laptop, and cables. The longer version of that argument is ShapeScale vs Styku.

Fit3D now calls the ProScanner a legacy system (2013 to 2023). New installs are pushed to SNAP: a Samsung Galaxy tablet on a stand, $100/mo, no turntable, weight typed in by hand. Same reporting backend, per Fit3D. Do not treat a tablet photogrammetry app as the ProScanner literature below.

ShapeScale is the other capture: a robotic arm / rotating sensor head moves around a still person and builds a photorealistic mesh. Composition is predicted from that entire 3D mesh by a deep-learning model trained on paired scans vs GE Lunar iDXA, not from an inch equation. Same class of optical 3D. Different method.

What ShapeScale reports

ShapeScale's measurement list is the same class of inches and volumes Fit3D sells, plus composition trained on iDXA. Circumferences in the app are neck, shoulders, chest, biceps, waist, hips, thighs, and calves. Volumes cover arms, torso (upper, lower, buttocks), and legs (thigh, calf). Compare, Body, Heatmaps, and the PDF use one combined arm volume; those views no longer split bicep and tricep.

Composition includes body fat and lean mass as percentages, fat mass in lb or kg, lean mass, lean mass index (LMI, lean mass relative to height), appendicular lean mass (ALM, lean tissue in the arms and legs), visceral fat as a mass in lb or kg, basal metabolic rate (BMR), body-mass index (BMI), and the 0-100 Health Score and Shape Score. Height is typed in, not scanned. There is no posture score.

On a new scan you can move where a girth is taken. Existing scans keep the original location.

Dr. Johnny Franco (Austin Plastic Surgeon):

"ShapeScale has truly revolutionized how we guide patients through their weight loss journey. It makes it so much easier for them to visualize their transformation and track progress, because it is no longer just about losing weight."

Source: ShapeScale vs Styku

Trina Barr (Innovaes Med Aesthetics):

"ShapeScale provides visible, tangible, absolutely undeniable metric of proof that was missing from our practice."

Source: ShapeScale vs Styku

Dr. Jake Deutsch (Oakwood Precision Medicine), who replaced a Styku: "the [ShapeScale] images are photographic vs Styku's is kind of like a grey blob." Fit3D's mannequin is the same product class.

Source: ShapeScale vs Styku

How the walkthrough actually runs

Staff create the client in the app. Menu → Clients → New Client → Create. That is a managed account. The client does not have to set up an account themselves.

If you share or invite, the client also gets the iOS app. Both sides can see the scans. Staff work from the Clients tab on an iPhone or iPad. Exit Client means staff are inside that client profile. Android clients stay managed and get the PDF.

The first scan uses a QR code from the app. After that, ShapeScale uses body and face recognition. The client steps on while the device shows rainbow lights. No app. No buttons. That works across even hundreds of users. Staff do not pick the right profile or stand there capturing return scans.

That is the gap versus a Fit3D ProScanner. On the ProScanner the person holds the handles, the plate turns, and staff run the laptop. ShapeScale return scans are self-serve.

The consult is the Compare view: photoreal 3D before and after, plus the heatmap and the PDF. That is the visual. AI guidance on goals is in the app. That saves staff time in consultations and helps personalize treatment plans.

Photoreal is the visual product. Staff and clients can export before-and-after images and videos from Export Image, plus that PDF. People use those for social posts and testimonials, with consent. Fit3D hands you a mannequin. This is the photograph.

What the papers say

A 2017 Fit3D blog post said body-fat predictions were "within 5% for 90.4% of scans." That is a vendor post about their v4.0 algorithm, not a paper. We are not using it as accuracy.

The March 2026 white paper compared ShapeScale's current composition model to a GE Lunar iDXA. Fit3D was not in that study. We do not have a head-to-head Fit3D number of our own.

Independent groups have published on the ProScanner. RMSE is root-mean-square error, the typical size of a miss. R2 says how tightly the estimate tracks the reference. Lower RMSE and higher R2 are better. Fat-free mass (FFM) is everything that is not fat. Those papers often use researcher equations, a 4-compartment model, or a Hologic table. That is not the same reference as ShapeScale's composition model, which was trained and tested against a GE Lunar iDXA. GE calls iDXA its research-grade, most advanced whole-body scanner. Wong used a Hologic Discovery/A. GE and Hologic tables do not print the same fat and lean numbers (JBMR Plus 2024), so those papers are not a head-to-head with the white paper. They are also not always the commercial printout a clinic or gym hands the member. Tinsley 2020 put Fit3D and Styku in the same four-scanner study. Same error band.

What those papers actually say:

• Ng et al., European Journal of Clinical Nutrition 2016. Fit3D Proscanner, n=39 adults. The authors built their own equations from 3D anthropometrics vs DXA. Fat mass R2 0.95, RMSE 5.3 lb (2.4 kg). Fat-free mass R2 0.96, RMSE 4.9 lb (2.2 kg). Landmark placement biased the girths. Fit3D later wrote that they "revised the algorithm a bit from here."
• Tinsley et al., Clinical Nutrition 2020. Commercial FIT3D, Naked Labs, Size Stream, and Styku vs a 4-compartment model, n=171. Across those four scanners, body-fat RMSE was 3.7-6.1 percentage points. Fat and fat-free mass RMSE 6.2-10.1 lb (2.8-4.6 kg). Three of the four met a 5% equivalence test; all four showed proportional bias. The abstract does not isolate a Fit3D-only MAE. We are not inventing one.
• Wong et al., American Journal of Clinical Nutrition 2023. Fit3D ProScanner v4.x vs a Hologic Discovery/A, n=133, 3-23 weeks. Meshes were reposed in Meshcapade and run through published shape equations, not the commercial Fit3D fat number. Change in fat mass tracked (women R2 0.86, RMSE 4.4 lb / 1.98 kg; men R2 0.75, RMSE 5.1 lb / 2.31 kg). Change in %fat did not: R2 0.23 in women and 0.25 in men.

A 4-week GLP-1 or contouring change often sits inside that noise. If the number can move more than the treatment, staff stop using it as proof.

That white paper (optical 3D vs GE Lunar iDXA, n=1,000+ adults, no participant overlap across train / validation / test):

• Body fat: MAE 1.89 pp, R2 0.91
• Lean mass (abstract): MAE 3.3 lb (1.49 kg), R2 0.98
• Appendicular lean mass: MAE 2.1 lb (0.97 kg), R2 0.96

The study's limitations are clothing, hair, and stance, not hydration. Rotating-platform 3D (Fit3D, Styku) was not in that study. Same note as the best-scanner guide.

Bioelectrical impedance (BIA) is a different method. InBody and Evolt 360 estimate composition from resistance. Do not treat a Fit3D avatar as an InBody column.

App (checked 16 Aug 2026)

• Fit3D: No consumer iOS listing found. SNAP is an operator Android-tablet app (16 reviews observed). Results live in Fit3D's web dashboard.
• ShapeScale: 4.0 / 5 App Store (43), iOS only. 3D on iPhone or iPad during the consult. Android is not available.

Protocol sensitivity

• Fit3D ProScanner: Stance plus platform rotation. Clothing. Handles on. Fit3D says implants move the composition result, and that very lean athletes are over-read.
• ShapeScale: Clothing, hair, stance (white-paper limitations)

When Fit3D is enough

Keep a ProScanner if you already run it (common on gym floors), you sell posture or body-shape scores, and you will only use composition as a trend under a locked stance.

Do not buy another turntable, or a tablet that outputs the same inch-based fat number, if the job is photoreal proof plus girths and volumes in a gym, studio, plastics, GLP-1, or body-contouring consult.

Book a ShapeScale demo

FAQ

Is Fit3D a DXA scan?

No, and neither is ShapeScale. Both are optical 3D. Fit3D estimates composition with a deterministic equation from inches and surface volume (Fit3D FAQ: correlating anthropometrics to DXA readings). ShapeScale predicts composition from the full 3D mesh with a deep-learning model trained on paired scans vs GE Lunar iDXA. The scan is the outside of the body. DXA is a different method.

How does Fit3D get a body-fat number?

From the inches and surface volume, not from tissue. Fit3D says the algorithm is "primarily volumetric/anthropometric based." Two people with the same waist can get the same fat %. That is the same class as Styku, not BIA and not DXA. ShapeScale does not use that equation.

Is SNAP the same as the ProScanner?

No. The ProScanner is a rotating platform in front of a camera tower. SNAP is a Samsung Galaxy tablet on a stand, $100/mo, no turntable, weight typed in by hand. Fit3D says they share a reporting backend. We did not find a SNAP-vs-DXA paper. Do not treat SNAP as the ProScanner studies on this page.

Do Fit3D and ShapeScale both report inches and volumes?

Yes. Both report regional circumferences and regional volumes. ShapeScale lists neck, shoulders, chest, biceps, waist, hips, thighs, and calves, plus volumes for arms, torso, and legs. Fit3D is not the only scanner with girths. The gap is the walkthrough: mannequin on a laptop vs photoreal mesh and heatmaps on iPhone or iPad.

Does ShapeScale do posture analysis?

No. Fit3D reports tilt, shift, and ProScanner balance, often as a separate report. ShapeScale has no posture score. Keep a ProScanner if posture is the product you sell.

Who should keep a Fit3D ProScanner?

A clinic or gym that already runs one, sells posture or body-shape scores, and will only use the fat number as a trend under a locked stance. Do not buy another turntable if the job is photoreal proof in a plastics, GLP-1, or contouring consult.

Who should buy ShapeScale instead?

A gym or studio, med spa, plastics practice, wellness clinic, or GLP-1 program that needs a still-person photoreal mesh, girths, volumes, and composition trained on GE Lunar iDXA, shown on iPhone or iPad during the consult or floor walkthrough, plus a PDF they can take.

What do the papers actually say about Fit3D accuracy?

Fit3D's 2017 blog said "within 5% for 90.4% of scans." That is a vendor post, not a paper. Independent ProScanner work: Ng 2016 n=39, fat-mass RMSE 5.3 lb (2.4 kg); Tinsley 2020 four scanners including FIT3D, body-fat RMSE 3.7-6.1 pp vs a 4-compartment model, no Fit3D-only MAE in the abstract; Wong 2023 %fat change R2 0.23-0.25 vs Hologic. ShapeScale composition model vs GE Lunar iDXA: body-fat MAE 1.89 pp, R2 0.91. Fit3D was not in that study.

Can staff walk a patient through results in the room?

On Fit3D, results live in a web dashboard. Posture is often ready a day later. On ShapeScale, photoreal 3D, girths, volumes, and heatmaps are on iPhone or iPad during the consult. Staff can export a PDF of one scan or a two-scan compare. The app is iOS only. Android is not available.

Does a ShapeScale still on this page mean the consult is iPad-only?

No. The app runs on iPhone or iPad. The girth-overlay, heatmap, dashboard, graph, and Compare stills on this page are iPhone screenshots. Height is typed in, not scanned.

What do ALM, LMI, and FFM mean?

Appendicular lean mass (ALM) is lean tissue in the arms and legs, where most skeletal muscle sits. A whole-body lean or fat-free number is everything that is not fat: muscle, organs, water, and gut contents, and it can move when someone drinks or eats. ALM is the number a GLP-1 or training program actually wants: did the limbs keep the muscle. Lean mass index (LMI) is lean mass relative to height, similar to BMI. Fat-free mass (FFM) is everything that is not fat. ShapeScale reports ALM and LMI today. Papers on Fit3D often report FFM. Fit3D does not report ALM.

What do MAE and RMSE mean?

Mean absolute error (MAE) is the average size of a miss versus the reference (here, GE Lunar iDXA in the white paper). Root-mean-square error (RMSE) is a similar miss size that weights larger misses more. Lower is better. R2 says how tightly the estimate tracks the reference. Higher is better.

How much floor space does each scanner need?

Fit3D's ProScanner help asks for a clear floor of 78 by 62 inches, plus a 1-foot keep-clear. ShapeScale needs 5 ft of rotation clearance, 6 by 6 ft recommended, and 8 ft of ceiling. The person stands still on a 2-foot base. The sensor moves.

Can both the provider and the client use the app?

Yes, if staff share or invite. Both sides see the scans on iOS. Staff work from the Clients tab on their iPhone or iPad. The app is iOS only. Android is not available.

Does the client have to create their own account?

No. Staff can create a managed account from Menu → Clients → New Client → Create. Share if you also want the client on the iOS app. HIPAA-private clients still create their own private account.

Do staff have to tap anything to scan a return client?

No. The first scan uses a QR code from the app. After that, the client steps on while the device shows rainbow lights. Body and face recognition start the scan. No app. No buttons. Staff do not stand there for return captures the way they do on a Fit3D ProScanner (handles, plate, laptop).

Can we export images or videos for marketing?

Yes. Export Image makes before-and-after images and videos. The existing PDF covers one scan or a two-scan compare. People share those for social posts and testimonials, with consent.

Does recognition work with a lot of clients?

Yes — after the first QR scan, face and body recognition starts the return scan. That works across even hundreds of users. No app. No buttons.

Does the app help personalize a treatment plan?

Yes. AI guidance on goals is in the app. That is consult time you do not have to spend walking someone through a plan from scratch. Compare + heatmap + PDF are still the visual.

Sources

• ShapeScale composition white paper, March 2026 (BF% MAE 1.89 pp, R2 0.91; lean MAE 3.3 lb / 1.49 kg, R2 0.98; ALM MAE 2.1 lb / 0.97 kg, R2 0.96; n=1,000+ vs GE Lunar iDXA)
• ShapeScale measurement list (circumferences including chest; volumes; fat mass, LMI, ALM, VAT as mass)
• Fit3D, ProScanner described as legacy rotating-platform hardware; SNAP as current tablet product
• Fit3D, ProScanner = tower + rotating platform; SNAP $100/mo
• Fit3D, evolution of tower scanners (~$9,000 ProScanner)
• Fit3D homepage (4M+ scans; SNAP software-only)
• Fit3D Body Fat FAQ (composition from anthropometrics; Ng cited; algorithm revised after Ng)
• Fit3D posture reports (tilt, shift, ProScanner balance; processed separately from the scan report)
• Fit3D how to review client scans (web dashboard)
• Fit3D ProScanner dimensions and spacing (78 × 62 in required clear floor; 38 in turntable; 65.25 in tower-to-plate; 1-foot keep-clear)
• Fit3D 2017 vendor post (v4.0 "within 5% 90.4% of the time", vendor blog, not a paper; not used as our MAE)
• Ng et al., Eur J Clin Nutr 2016 (Fit3D Proscanner, n=39, FM RMSE 2.4 kg, FFM RMSE 2.2 kg)
• Tinsley et al., Clin Nutr 2020 (FIT3D + three other 3DO vs 4C, n=171, BF% RMSE 3.7-6.1 pp)
• Fit3D write-up of Tinsley (PMID 32113641)
• Wong et al., Am J Clin Nutr 2023 (Fit3D ProScanner vs Hologic Discovery/A, n=133, %fat change R2 0.23 / 0.25)
• SNAP by Fit3D, Google Play (operator tablet app, 16 reviews, 16 Aug 2026)
• ShapeScale App Store (4.0 / 5, 43 ratings, iOS, 16 Aug 2026)
• ShapeScale vs Styku (Franco, Barr, Deutsch; Styku as the other rotating-platform comparison)
• Naake et al., 3DBODY.TECH 2024 (turntable-class subject motion up to ~2 cm on a Scaneca turntable, not a Fit3D or Styku unit)
• Evolt 360 for clinics
• InBody scan accuracy for a practice
• Best body scanner for a clinic
• GE HealthCare, Lunar iDXA (research-grade / most advanced GE whole-body DXA)
• Ghalla-Marti et al., JBMR Plus 2024 (GE Lunar iDXA vs Hologic Horizon A: systematic fat/lean differences; not interchangeable)
• ShapeScale PDF reports (export one scan or two-scan compare; share from the app)
• ShapeScale managed accounts (staff create the client in the app; optional share so the client gets iOS)
• ShapeScale Clients tab (staff iPhone or iPad; Exit Client leaves the client profile)
• ShapeScale first scan (QR from the app; later scans use recognition)
• ShapeScale LED colors (rainbow lights mean ready to scan)
• ShapeScale Compare view (photoreal 3D side by side)
• ShapeScale dashboard (scores, composition, measurement tiles)
• ShapeScale on client results (photoreal 3D and heatmaps; share side-by-side transformations)
• ShapeScale product (Martin Kessler, 18 Aug 2026)