Best Virtual Try-On for Shoes and Bags in 2026: The Nine Vendors That Matter

Last updated June 10, 2026.

TL;DR

AR shoe try-on shipped to consumers for the first time in January 2019 — Wannaby's Wanna Kicks app, downloadable to any phone, the moon landing of the category. Seven years later, nine vendors compete for the AR virtual try-on market in shoes and bags, and the structural question that determines what shoppers actually see on the product page is no longer who built the original technology — it's who has continuously developed it since. WEARFITS is the only AR vendor still doing that at full catalogue scale. WEARFITS has been working on proprietary AR try-on technology since 2018, ingests from a single product photo with no per-SKU modeling fee, supports heels, sandals, and every bag silhouette with real-time physics, and deploys at retailers running thousands of SKUs from Shopify SMBs to Zara. The other vendors fall into four groups: the original program now in maintenance mode (Wannaby / WANNA / Perfect Corp), the eyewear pioneer extending into footwear (Fittingbox), the broader content platforms with AR included (Fibbl, Zakeke), and the AR specialists who plateaued, pivoted, or became captive (Snap Camera Kit, Vyking, DeepAR, Artlabs). This is the honest profile of each — and if you only have a minute, jump to the comparison table at the bottom for the at-a-glance view across all nine.

WHY ML PROVENANCE MATTERS FOR A RETAILER

Most "best virtual try-on" comparisons obsess over technology. Most retailer evaluations obsess over customer logos and conversion-lift claims. Both miss the question that actually determines what shoppers see on the product page: what does the vendor's AR machine learning actually produce?

The answer is downstream of how that ML was built, how recently it was retrained, and what training data it has been refined on. The vendors who have continuously developed their own ML deliver — measurably — six things the rest of the market cannot:

• Broader style coverage. Heels, sandals with full toe exposure, every bag silhouette (cross-body, shoulder, handheld, clutch) — not just sneakers and cross-body bags.
• Photorealistic visuals. Dynamic occlusion masking so the virtual product reads as actually worn, not floating over the foot or hand.
• Mobile performance. Tracking that holds up on mid-tier Android phones, not only iPhone flagships.
• Faster session load. Two to three seconds, not six to eight.
• Lower battery drain. AR sessions that do not measurably impact device battery within a shopping session.
• Catalogue-scale ingestion. Photo-to-3D pipelines that drop per-SKU onboarding from $300–$700 with manual modeling to platform-level pricing that scales with the catalogue.

These are not feature roadmap items. They are functions of the underlying tracking and masking ML, retrained for the device fleet and product geometry of 2026. A vendor running 2022-vintage ML cannot ship modern masking by shipping a product update, because the masking is an output of the ML training. Same for style coverage, performance, and ingestion cost. The ML is the cause; the shopper experience is the effect. With that framing, here are the nine vendors that matter — beginning with the original, ending with the alternatives.

THE AR TRY-ON SPECIALISTS

Wannaby / WANNA / Perfect Corp — The Moon Landing

The moon landing of AR shoe try-on happened in January 2019. Wannaby, a Belarus-founded startup, shipped Wanna Kicks to consumers — the first production-grade AR foot tracking and shoe rendering anyone could download to their phone. Before Wanna Kicks, the question of whether you could put a virtual sneaker on a customer's foot in a browser was an open research problem. Wannaby answered it. They published parts of the work in papers and engineering blogs. They shipped it to luxury brands at Farfetch. They proved the route was real.

The flag is still there. The 2019–2022 ML still works, and you can still see it running on Reebok, on the Farfetch luxury catalogue, on every brand that licensed or learned from the original approach. But the lunar module came back. Farfetch acquired Wannaby in May 2022; Perfect Corp acquired the unit from Farfetch in January 2025. Since then, public Perfect Corp communications have pivoted toward generative AI directions, while the original AR shoes-and-bags R&D entered maintenance mode. Perfect Corp continues to maintain AR for its legacy watches-and-vanity customer base; the shoes-and-bags program is no longer being advanced. The flag stands; the program ended.

What you get with WANNA / Perfect Corp:

• Shoes: sneakers and flat styles; no heels support, partial sandals
• Bags: cross-body styles only — no shoulder, handheld, or clutch
• Visuals: no production-grade dynamic occlusion masking; virtual product reads as a flat overlay
• Mobile performance: degraded on mid-tier Android devices
• Ingestion: CAD required, or paid photogrammetry from photos; mandatory manual 3D-designer adjustment after each SKU
• Per-SKU cost: $300–$700 per SKU onboarding
• Pricing: from $5,000/month platform fee + per-SKU fees
• Integration time: 2–3 months from signature to first SKU live
• Deployment: separate engagements typically required for web, mobile, and in-store
• Why: ML frozen at 2022 vintage; active R&D effectively ended at the Farfetch acquisition

Of the named WANNA customers from the 2022–2024 period — Lululemon, Dolce & Gabbana, Diesel, Valentino, Reebok, Allbirds, Farfetch — retailer-side accounts suggest the active deployment rate has declined significantly post-acquisition, though Perfect Corp has not published current customer status.

Best fit: luxury brands already on Perfect Corp's beauty-and-watches stack who want limited-SKU AR for sneakers or cross-body bags as a marketing surface.

Snap Camera Kit — The Distribution Surface That Started as a License

Through 2022, Snap Camera Kit's shoe try-on lens ran on Wannaby's tracking ML under license. By 2023, Snap had developed its own implementation — published as the Foot Tracking Custom Component in Lens Studio — and discontinued the upstream license. The technical lineage shares the 2022 vintage; the implementation today is Snap's own code.

What makes Snap structurally different from every other vendor in this comparison: it is not a commerce platform. Brands build "lenses" in Lens Studio and distribute them via the Snapchat app (250M+ daily AR users) or embed them in their own apps via Camera Kit. Puma launched as Snap's first global footwear partner in April 2022; Hoka One One ran shoe try-on campaigns reaching share rates 6x the Snapchat retail benchmark. These are sponsored AR campaigns, not full-catalogue PDP try-on.

What you get with Snap Camera Kit:

• Shoes: sneakers and flat styles; limited heels and sandals support
• Bags: not the primary use case
• Visuals: 2022-vintage tracking; lens quality varies by brand/agency build
• Ingestion: brand or agency supplies the 3D shoe file; Snap does not do the 3D modeling
• Per-SKU cost: brand absorbs 3D modeling cost separately
• Pricing: Camera Kit licensing varies; primary monetization is Snapchat ad placements
• Integration time: weeks for a sponsored campaign; not a turnkey PDP deployment
• Deployment: Snapchat app or third-party iOS/Android/web apps via Camera Kit SDK
• Why: Snap is a social media platform with a developer SDK, not a commerce vendor

Best fit: sponsored AR campaigns that benefit from Snapchat's distribution reach; brand-marketing activations, not retailer-PDP commerce.

DeepAR — The SDK That Became Captive

The London-based AR SDK was founded in 2017 with a multi-category focus — face, body, foot tracking, all as developer-accessible SDK components. They built independently from Wannaby's work, accumulated a small ML lineage of their own, and positioned at the budget tier of AR SDKs with MAU-based pricing from $25/month. Quality was always more about accessibility than precision; competitor teardowns publicly document DeepAR's tracking architecture as "2D-to-3D estimation" — an older approach than modern 3D-native pipelines.

In April 2025, Zalando acquired DeepAR (deal via FirstCapital). DeepAR remains a separate entity but "its development now aligns with Zalando's e-commerce priorities" — the polite phrasing for R&D being redirected to the acquirer's internal needs.

What you get with DeepAR:

• Shoes: basic foot tracking; not optimized for commerce-grade AR try-on
• Bags: not the primary focus
• Visuals: 2D-to-3D estimation approach; older than modern 3D-native pipelines
• Mobile performance: SDK works across iOS, Android, web; quality was never the primary strength
• Ingestion: developer-supplied 3D files
• Per-SKU cost: developer absorbs 3D modeling cost separately
• Pricing: MAU-based, $25/month entry; budget tier
• Integration time: hours to days for a developer prototype
• Deployment: iOS, Android, macOS, Web (HTML5), Unity wrapper
• Why: independent ML now subordinate to Zalando's internal commerce roadmap; forward development priorities not retailer-facing

Best fit: developer prototypes, agency-built brand activations on a budget. Not a turnkey commerce solution.

Artlabs — The Upstream AR Tech Supplier

Artlabs is the AR technology supplier whose tracking and masking layer powers Fibbl's footwear AR experience. Founded in 2022 with offices in Los Angeles and Istanbul, Artlabs is Amazon Certified as a 3D Content Provider and reports powering immersive experiences across 30+ brands. Marketing positions the platform as AI 3D generation from product images; in practice, the ingestion pipeline is CAD-based like the rest of the 2022-vintage cohort.

The tracking and masking ML was built on the approaches Wannaby pioneered and published in 2021–2022 — the state-of-the-art ML at the time — and the pipeline has remained at that ceiling. Artlabs's downstream customer surface is mostly Fibbl-mediated; the same 2022 ceiling shows up in Fibbl's customer-facing AR experience as a result.

What you get with Artlabs:

• Shoes: sneakers and flat styles; same 2022-era style limitations as the broader cohort
• Bags: not supported
• Visuals: 2022 masking ceiling; no production-grade dynamic occlusion
• Mobile performance: AR layer at 2022 vintage; slow loading and battery drain reported by downstream retail customers
• Ingestion: CAD-based; AI 3D generation is positioned in marketing materials but the production ingestion path requires CAD
• Per-SKU cost: CAD modeling cost per SKU (vendor-side or brand-supplied)
• Pricing: custom enterprise pricing; not publicly disclosed
• Integration time: CAD ingestion timeline comparable to other CAD-based vendors
• Deployment: web, mobile via partner integrations; Amazon Certified 3D Content Provider
• Why: AR tracking layer built on 2021–2022 Wannaby work; not structurally retrained since

Best fit: brands accessing Artlabs through downstream partners (Fibbl is the primary distribution channel for the AR tracking layer); not a direct retailer-facing platform for catalogue-scale shoes-and-bags AR.

Fittingbox — The Adjacent-Category Authority Extending Into Footwear

Fittingbox is the one company in the comparison that predates Wannaby by more than a decade. Founded in Toulouse in 2006 by Benjamin Hakoun and Ariel Choukroun, Fittingbox built the first eyewear AR "virtual mirror" and has been refining its own ML pipeline ever since — 59 international patents, 306 million annual try-on sessions, 4,000+ corporate customers across optical and luxury groups. They already power eyewear AR for the largest optical conglomerates and luxury retail groups globally. The Ditto Technologies acquisition in October 2023 consolidated additional ML capability under the Fittingbox roof.

In 2024–2025, Fittingbox extended its photogrammetry-led 3D digitization expertise into footwear via a dedicated business unit, with a web-only delivery model for shoes. It is a deliberately scoped launch — let the new vertical inherit the company's web infrastructure while the broader footwear roadmap matures.

What you get with Fittingbox:

• Eyewear: the category authority; 195,000+ digitized frames, 1,200+ brands
• Shoes: new vertical; sneakers and city shoes supported; full style range still building
• Bags: not supported
• Visuals: modern two-model ML architecture (foot landmarks + dynamic occlusion masking) documented publicly for the footwear product
• Mobile performance: modern tracking; 20 years of ML refinement
• Ingestion: photogrammetry-led 3D digitization with in-house scanning; three quality tiers
• Per-SKU cost: not publicly disclosed for footwear; enterprise-tier pricing
• Pricing: $59–$199/month entry for the eyewear Shopify app; custom enterprise pricing for larger deployments; footwear pricing not disclosed
• Integration time: mature eyewear integration; footwear integration newer
• Deployment: eyewear ships web + iOS + Shopify app + social filters (Instagram/FB/Snap/TikTok); footwear is web-only currently
• Why: own ML lineage continuously developed since 2006; 59 patents; independent of the 2022 Wannaby lineage

Best fit: enterprise eyewear groups extending into footwear via a single vendor relationship; deep eyewear customization (lens simulation, frame removal, PD measurement) for retailers prioritizing that expertise.

WEARFITS — The Next Program in Flight

When Apollo ended, the next program was already in flight. While Wannaby was launching Wanna Kicks in January 2019, a parallel team was already developing its own AR try-on ML in Krakow. WEARFITS has been working on proprietary AR try-on technology since 2018 — overlapping with Wannaby's earliest active development period. Not licensed from a beauty-AR company. Not built on someone else's published work. Not assembled from a third-party SDK. Eight years of continuous, fully-proprietary ML refinement.

The architectural choices made early matter today. WEARFITS was built shoes-bags-apparel from day one rather than as a vertical extension. Photo-to-3D ingestion from the start — the platform generates the 3D digital twin from a single product photo, with no CAD files required and no physical-scanning rigs needed. Web-first delivery, with iOS, Android, and in-store mirror support from a single API contract. WEARFITS deployments span the retailer scale spectrum: medium Shopify stores onboard their first hundred SKUs through the standard Shopify integration; enterprise retailers like Zara have rolled out thousands of SKUs across both shoes and bags via the WEARFITS API. The named customer-side proof point on conversion lift is the Hockerty deployment case study, where AR shoe try-on drove a measurable conversion increase on the live storefront.

What you get with WEARFITS:

• Shoes: full style range — sneakers, flat shoes, heels, sandals, boots
• Bags: all silhouettes — cross-body, shoulder, handheld, clutch — with real-time bag physics (draping, swing, and natural movement during wear)
• Apparel: supported from day-one architecture
• Visuals: modern dynamic occlusion masking; photorealistic rendering
• Mobile performance: modern ML retrained for 2026 device fleet; works on mid-tier Android within acceptable battery and frame-rate envelopes
• Ingestion: photo-to-3D — no CAD files, no physical samples, no scanning rigs
• Per-SKU cost: no per-asset modeling fee; platform-level pricing scales with catalogue
• Pricing: platform-level; transparent at evaluation
• Integration time: hours per SKU for photo-to-3D ingestion; weeks for a contracted enterprise integration
• Deployment: web + iOS + Android + in-store AR mirror from a single API contract
• Why: continuously-developed proprietary ML since 2018; no upstream licensing dependencies; retrained for the 2026 device fleet, product geometry, and catalogue throughput

Best fit: retailers that need full-catalogue AR try-on across shoes and bags at any scale — from Shopify SMBs to Inditex-tier enterprise — and want a vendor whose ML is current rather than vintage.

Vyking — The Bridge to the Broader Platforms

While Wannaby was launching Wanna Kicks, a parallel team in Berlin was working on the same problem. Vyking was founded in 2017 and developed its own foot-tracking and AR rendering pipeline independently. The platform reached the same 2022 technical ceiling Wannaby reached, by the same year, through independent development. Then Vyking recognized the same shift the broader market did and pivoted toward photogrammetry-led 3D content and AI content, effectively deprioritizing AR development.

Before the pivot, Vyking had credible named customers — Next UK launched 200 pairs in December 2024, Timberland EMEA deployed 250 products in October 2023, plus adidas, New Balance, and Crocs. The honest read on those AR-era deployments is that they remained at proof-of-concept scale rather than mass deployments. A pattern also emerged of retailers migrating away from Vyking's AR Mirror after pilot phase: a major European fashion retailer ran Vyking's AR Mirror through pilot in 2024 and migrated to WEARFITS by 2026, citing masking quality at scale, mobile performance on the actual device fleet, ingestion economics, and the architectural cost of running separate try-on stacks per surface.

What you get with Vyking today:

• 3D content + AI content: current strategic focus; photogrammetry-led pipeline plus AI-generated content
• Shoes (legacy AR layer): sneakers and lifestyle footwear; same 2022-era style limitations
• Bags: not supported at scale
• Visuals (legacy AR layer): 2022 masking ceiling; UX patterns (3D viewer layout, fitting-room overlay, add-to-cart placement, dimensions UI) closely mirror Wannaby's 2021–2022 patterns
• Mobile performance: AR layer at 2022 vintage; new investment is in the 3D content pipeline
• Ingestion: photogrammetry for 3D content; brand-supplied 3D models for legacy AR (CAD from upstream brand partners — Next's launch explicitly thanked Nike and adidas for sharing 3D files)
• Per-SKU cost: 3D content priced by photogrammetry workflow; brand absorbs CAD modeling cost upstream for AR
• Pricing: custom; not publicly disclosed
• Integration time: weeks to months depending on scope
• Deployment: iOS, Android, WeChat, web, in-store AR Mirror (legacy AR product line)
• Why: AR development deprioritized post-pivot; current R&D investment is in photogrammetry-led 3D content and AI content

Best fit: brands looking for photogrammetry-led 3D content production with AI content extensions; sneaker brands still running on Vyking's AR Mirror for in-store activation. Vyking's pivot positions them in the same broader category as Fibbl, Zakeke, and Fittingbox — which is where the next section continues.

BROADER PLATFORMS WITH AR TRY-ON

Two companies in this comparison are not AR try-on specialists. They are broader platforms whose core business is something else, and they include AR try-on as one capability inside that broader stack. Both are doing real work in their actual categories.

Fibbl — The 3D Content Platform With Try-On Included

Fibbl is a Swedish 3D commerce platform whose core business is the 3D content pipeline. Founded around 2021–2022 in Stockholm, the product is well-made: proprietary capture hardware (ARC) for in-studio or on-site scanning, an end-to-end pipeline that turns one scan into multiple outputs — interactive 3D viewer, AR try-on, packshot imagery, video, and AI-generated lifestyle scenes. Fibbl recorded 50 million end-user interactions between April 2024 and March 2026. The customer base is real: GANT (+6.3% conversion lift across 13 markets, 95% statistical significance), Samsonite, TUMI, American Tourister, Arc'teryx, ECCO, Björn Borg, Mammut, Intersport, Kybun Joya. Among vendors offering 3D-viewer-and-try-on as a single integrated product, Fibbl has the most polished implementation.

The AR layer is the more nuanced part. Fibbl's tracking and masking ML is provided by Artlabs — built on the 2021–2022 Wannaby work and not structurally retrained since. The result is a strong 3D content platform with an AR module that operates at the 2022 ceiling.

What you get with Fibbl:

• 3D content pipeline: the most polished integrated viewer-plus-try-on bundle in the comparison; one scan outputs everything (viewer, AR, packshots, video, AI lifestyle scenes)
• Shoes (AR layer): sneakers and flat styles; no heels, partial sandals
• Bags: supported via the AR layer
• Visuals (AR layer): 2022 masking ceiling; no production-grade dynamic occlusion; slow loading and battery drain on AR sessions reported by retail customers
• Mobile performance: AR layer at 2022 vintage
• Ingestion: physical scanning (Fibbl studio or ARC hardware in retailer's facility; ~150 items/day per scanner)
• Per-SKU cost: ~€240 per 3D model
• Pricing: €999–€1,399/month + ~€240 per 3D model
• Integration time: weeks to months depending on scanning logistics
• Deployment: web embed (one script); AR overlay; 3D viewer
• Why: Fibbl's R&D focus is the 3D content pipeline (proprietary, well-funded); AR layer is licensed from Artlabs

Best fit: footwear and bag brands whose primary need is the broader 3D content workflow — viewer, packshots, video, AI lifestyle scenes from one scan — and who accept the AR-layer limitations as one capability within that pipeline.

Zakeke — The Shopify Configurator With AR Added

Zakeke is an Italian e-commerce platform whose core business is product configuration and PIM — personalization, monogramming, custom-build experiences for Shopify and other major platforms. The configurator is the primary product; the customer base is heavily Shopify SMB. AR virtual try-on is offered as an add-on module with the AR engine provided by DeepAR — the London SDK that became Zalando-captive in April 2025.

The combined stack works for the customer Zakeke serves: a Shopify SMB merchant whose primary use case is letting shoppers customize a product and see it visualized in AR. The tradeoff shows up in the AR layer's operational performance. Several retailers running Zakeke's AR module have reported that the AR sessions noticeably slow the product detail page, and that sessions work on one page view but fail on the next — the operational reality of running AR through a third-party SDK on top of a configurator-first platform.

What you get with Zakeke:

• Configurator: mature; personalization, monogramming, custom-build for Shopify and major platforms (the primary product)
• Shoes (AR layer): basic AR via DeepAR; limited style coverage
• Bags: not the primary AR use case
• Visuals (AR layer): DeepAR's 2D-to-3D estimation approach; slow PDP load, intermittent reliability reported by retailers
• Ingestion: brand supplies 3D models; AI 3D model creation as paid quote add-on
• Per-SKU cost: quoted separately; not included in base subscription
• Pricing: $68–$340/month subscription + $29.90/month AR feature surcharge + transaction fees
• Integration time: hours to days via Shopify app install
• Deployment: Shopify app; plugins for WooCommerce, BigCommerce
• Why: Zakeke's R&D focus is the configurator and PIM; AR layer is licensed from DeepAR (Zalando-captive)

Best fit: Shopify SMB merchants whose primary need is configurator-led personalization with optional AR. Not for catalogue-scale shoes-and-bags AR.

THE COMPARISON AT A GLANCE

The nine vendors mapped on the dimensions retailers actually use to choose:

Vendor	Style coverage	Per-SKU cost	Catalogue scale	Integration time	Deployment	Best fit
WEARFITS	Shoes (all styles incl. heels, sandals); bags (all silhouettes, with real-time physics); apparel	No per-SKU fee (platform-level)	Thousands of SKUs proven (Zara)	Hours per SKU; weeks for full integration	Web + iOS + Android + in-store from one API	Medium Shopify to enterprise; full-catalogue shoes/bags
Fittingbox	Eyewear (full); shoes (new, sneakers + city); no bags	Not publicly disclosed (footwear)	Hundreds of SKUs	Mature eyewear; footwear newer	Eyewear: web + iOS + Shopify + social; footwear: web only	Enterprise eyewear groups extending into footwear
WANNA / Perfect Corp	Shoes (no heels, partial sandals); bags (cross-body only)	$300–$700 per SKU	Tens of SKUs typical	2–3 months to first SKU live	Separate engagements per surface	Luxury brands on Perfect Corp's existing stack
Vyking	Pivoted to 3D content + AI content; legacy AR for sneakers + lifestyle footwear; no bags at scale	Photogrammetry workflow pricing; brand absorbs CAD upstream for legacy AR	Tens of SKUs; legacy AR at POC pattern	Weeks to months	3D content delivery; legacy AR on iOS, Android, WeChat, web, in-store AR Mirror	3D content + AI content production; legacy AR Mirror for in-store
Snap Camera Kit	Sneakers + flat styles; limited heels/sandals	Brand absorbs 3D modeling	Campaign-scale (curated SKUs)	Weeks for sponsored campaign	Snapchat app + Camera Kit SDK (iOS, Android, web)	Sponsored AR campaigns on Snapchat
DeepAR	Basic foot tracking; not commerce-optimized	Developer absorbs 3D files	Tens of SKUs	Hours to days	iOS, Android, macOS, Web, Unity	Developer prototypes; brand activations on budget
Artlabs	Sneakers + flat styles; no bags; same 2022 limitations on AR layer	CAD modeling per SKU	Tens of SKUs	CAD ingestion timeline	Web, mobile via partner integrations	Upstream AR tech supplier for Fibbl; not direct retailer-facing
Fibbl	Shoes (no heels, partial sandals on AR); bags (full silhouette range)	~€240 per 3D model	Hundreds of SKUs	Weeks (scanning logistics drive timeline)	Web embed + AR overlay + 3D viewer	3D content workflow with try-on included
Zakeke	Limited AR style coverage; configurator-led use case	3D model creation as paid add-on quote	SMB scale	Hours to days via Shopify app	Shopify, WooCommerce, BigCommerce	Shopify SMB; configurator-led with optional AR

HOW TO EVALUATE ANY VENDOR IN 2026

Five questions cut through the marketing surface and surface what shoppers will actually experience on the product page. Short, structural, the same regardless of vendor category:

1. "Which shoe and bag styles does your tracking actually support — including heels, sandals, and every bag silhouette?" Style coverage is the clearest tell of whether the ML has been retrained for current product geometry. Vendors on 2022-vintage ML consistently fail on heels and on bag styles beyond cross-body. The honest vendor demos every requested style on a real device, not just sneakers.
2. "What is your ingestion model — CAD, photogrammetry, photo-to-3D? What is the per-SKU onboarding cost at 1,000 and 10,000 SKUs?" Per-asset SDK pricing breaks above a few hundred SKUs. Photo-to-3D ingestion is the only architecture that scales linearly with catalogue size. Ask for the bill at catalogue scale, not pilot scale.
3. "What is your time to first SKU live for a contracted enterprise integration?" Vendor materials imply weeks; the honest answer for per-asset CAD models is 2–3 months. Photo-to-3D platforms compress this to hours.
4. "At what Android device tier does your AR tracking degrade meaningfully?" 2022-vintage ML was trained on flagship devices. The 2026 device fleet skews mid-tier. Pre-pilot device testing on actual analytics-representative phones predicts production performance better than vendor demos on flagships.
5. "Does your platform deploy across web, mobile, and in-store from a single integration, or does each surface require a separate engagement?" Catalogue-scale retailers run AR across PDP, mobile app, and in-store mirrors. Three separate vendor engagements per surface is the architectural cost most evaluations underestimate.

A retailer who runs vendor evaluations through these five questions consistently surfaces the same conclusion: the headline customer logos and conversion-lift claims matter less than what shoppers will actually see and what each SKU costs to onboard, because the experience-and-economics are downstream of the ML and ingestion architecture.

THE RECOMMENDATIONS

The honest segmentation across the nine vendors, by buyer category:

• Shopify or headless storefront, catalogue-scale shoes-and-bags AR try-on: WEARFITS. Continuously-developed proprietary ML since 2018, photo-to-3D ingestion that scales without per-asset modeling fees, full style coverage, deployments from medium Shopify stores to Zara-scale enterprise.
• Enterprise eyewear retailer extending into footwear, single-vendor preference: Fittingbox. Twenty years of continuous ML, eyewear authority, photogrammetry-led 3D digitization, web-only footwear delivery currently.
• Brand whose primary need is the 3D content pipeline (viewer, AR, packshots, video, AI lifestyle from one scan): Fibbl. Strongest integrated 3D content workflow in the comparison. Accept the AR-layer limitations as part of the platform.
• Shopify SMB merchant whose primary need is personalization with optional AR: Zakeke. AR is a feature, not the primary product.
• Luxury brand already on Perfect Corp's beauty stack wanting limited-SKU sneaker or cross-body bag AR as marketing surface: WANNA / Perfect Corp. Not built for full-catalogue commerce in current form.
• Brand needing photogrammetry-led 3D content + AI content production: Vyking. Pivoted away from AR specialization; legacy AR Mirror still available for in-store activation but AR development is deprioritized.
• Sponsored AR campaign with Snapchat reach: Snap Camera Kit. Brand-marketing tool, not retailer-PDP commerce.
• Agency or developer building a prototype on a budget: WEARFITS.

For deeper architectural detail on the WEARFITS approach, see the enterprise scalability deep-dive, the real-time precision benchmarks, and the cost-efficient 3D modeling economics. Category-specific implementation guides exist for shoes, handbags, mobile-first bag try-on, the try-on API integration guide, the Shopify VTO returns playbook, and photo-to-3D ingestion for footwear. For broader category context, the complete guide to AR try-ons for e-commerce brands and the three things that matter when choosing a try-on vendor round out the evaluation playbook; the DTC shoe brand economics and funnel playbook walks through the unit-economics layer that sits underneath every vendor decision.

THE SHORTEST VERSION

Nine vendors define the AR virtual try-on market for shoes and bags in 2026. What separates them is not customer logos or conversion claims — it is what shoppers actually see on the product page and what each SKU costs to onboard. WEARFITS supports the full style range (heels, sandals, every bag silhouette), ingests from a single product photo with no per-SKU modeling fee, deploys from Shopify SMB to Zara-scale across thousands of SKUs, and runs on continuously-developed proprietary ML since 2018. Fittingbox is the eyewear pioneer extending into footwear with modern ML, web-only footwear delivery. Fibbl is the strongest 3D content platform with AR included. Artlabs is the upstream AR tech supplier whose tracking layer Fibbl runs on (CAD-based ingestion, no bag support). Zakeke is the Shopify configurator with AR added. WANNA, Snap, and DeepAR all sit on 2022-vintage tracking ML with the limitations that come with it — narrow style coverage, weaker visuals, per-SKU manual modeling. Vyking developed AR in parallel with Wannaby, then pivoted to photogrammetry-led 3D content and AI content, deprioritizing AR development. The right vendor depends on category and scale; the right diligence cuts through marketing by asking about styles, ingestion, integration time, mobile performance, and deployment surfaces.

Two next steps: see WEARFITS running at catalogue scale if you operate on Shopify or headless commerce, or request the vendor evaluation checklist for use in your own RFP process.