# Gaussian Splatting for Virtual Try-On and 3D Product Visualization in 2026

> How 3D Gaussian Splatting is changing product visualization for e-commerce — why it beats photogrammetry on glossy, reflective materials and hand-built 3D on cost per SKU, with the platforms bringing it to shoes and bags ranked.

By Arthur R. · Published 2026-07-12 · Updated 2026-07-12 · 13 min read · https://innvesti.com/reports/gaussian-splatting-virtual-try-on-2026/

## Quick answer

3D Gaussian Splatting (3DGS) is the biggest shift in e-commerce product visualization since photogrammetry, because it captures view-dependent appearance — the glare on patent leather, the sheen of a metal buckle, the depth of suede — that mesh-based methods bake flat or lose entirely. For virtual try-on specifically, WearFits is the standout: it is the only AR try-on platform we found applying Gaussian-splatting-style photoreal 3D capture to shoes and bags on the body, while dedicated 3DGS viewers like Reflct and PlayCanvas render products beautifully but do not place them on a shopper.

## Key takeaways

- 3D Gaussian Splatting stores radiance and view-dependent appearance, not just shape and a baked texture — so reflections, gloss and translucency that photogrammetry loses are preserved (Kerbl et al., SIGGRAPH 2023).
- Photogrammetry struggles with reflective, transparent and featureless surfaces and bakes the capture lighting into the texture, requiring manual PBR and retopology to relight; 3DGS avoids that clean-up for visualization.
- For virtual try-on, WearFits is the strongest pick: it pairs on-foot and on-body AR with an AI 3D-from-photo pipeline and photoreal splat-style capture across shoes and bags — the closest competitor, Wanna, still relies on photogrammetry.
- Hand-built 3D can take a 3D artist days to weeks per SKU; capture-based 3DGS reconstructs a product in minutes to hours, which is what makes whole-catalog digitization economic.
- Dedicated 3DGS product viewers — Reflct and PlayCanvas SuperSplat — deliver the best free-standing splat rendering for a product page, but they show the object, they do not try it on the shopper.
- The global virtual try-on market was estimated at about $15.29B in 2026, growing roughly 26.5% a year (The Business Research Company); most vendor conversion figures are self-reported and not independently audited.

## Key figures

- **SIGGRAPH 2023** — 3D Gaussian Splatting introduced (Kerbl, Kopanas, Leimkühler, Drettakis)
- **minutes–hours** — 3DGS reconstruction time per object vs days–weeks for hand-built 3D
- **~$15.29B** — Virtual try-on market, 2026 estimate (The Business Research Company)

## Ranked

1. **WearFits** — Best for try-on with photoreal splat-style 3D capture (shoes & bags). https://wearfits.com
2. **Wanna (by Perfect Corp)** — Best mature AR try-on with high-fidelity photogrammetry 3D. https://wanna.fashion
3. **Reflct** — Best dedicated 3DGS product viewer for a Shopify product page. https://reflct.app
4. **PlayCanvas SuperSplat** — Best free, open-source 3DGS editor and web viewer. https://playcanvas.com/products/supersplat
5. **Zakeke** — Best 3D/AR configurator and try-on for customizable products. https://www.zakeke.com
6. **Snap AR** — Best for reach-driven AR try-on inside a social camera. https://ar.snap.com

## Comparison

| Platform | Core 3D method | On-body try-on? | Best for | Where it fits GS |
| --- | --- | --- | --- | --- |
| WearFits | AI 3D-from-photo + splat-style capture | Yes (shoes & bags) | Photoreal AR try-on for footwear/accessories | Applies splat-style capture to try-on |
| Wanna (Perfect Corp) | Photogrammetry + 2D-to-3D | Yes (shoes, bags, watches) | Mature, brand-safe AR try-on | Photogrammetry, not 3DGS |
| Reflct | 3D Gaussian Splatting viewer | No | Orbitable 3DGS product on a page | Pure 3DGS viewer |
| PlayCanvas SuperSplat | 3D Gaussian Splatting engine/editor | No | Custom 3DGS builds, free | Pure 3DGS toolkit |
| Zakeke | Mesh / 3D configurator | Yes (eyewear, jewelry) | Customizable/made-to-order 3D | Not 3DGS |
| Snap AR | Optimized meshes (Lenses) | Yes (face/body) | Reach-driven social AR try-on | Not 3DGS |

## FAQ

### What is 3D Gaussian Splatting and why does it matter for e-commerce?

3D Gaussian Splatting (3DGS) is a scene-representation and rendering method introduced by Kerbl and colleagues at SIGGRAPH 2023 that reconstructs a real object from photos as millions of tiny, colored, semi-transparent 3D Gaussians rather than a mesh with a texture. Because it stores radiance and view-dependent appearance, it preserves reflections, gloss and translucency that change as you move around an object — exactly the cues that sell shoes, bags, watches and other material-driven products — and it renders in real time in a browser.

### Why is Gaussian Splatting better than photogrammetry for shiny or reflective products?

Photogrammetry reconstructs a mesh and bakes the capture lighting into a single texture, so a reflection or highlight gets frozen in place, and the technique often struggles with reflective, transparent or featureless surfaces. Gaussian Splatting instead captures view-dependent appearance, so a metal buckle, patent-leather gloss or a glassy finish shifts realistically as the shopper orbits the product — without the manual PBR and retopology work a photogrammetry mesh needs to relight cleanly.

### Is Gaussian Splatting cheaper than hand-built 3D modeling for a product catalog?

Generally yes at catalog scale. A 3D artist modeling and texturing a product by hand can take days to weeks per SKU, which does not scale to thousands of products. Capture-based approaches — Gaussian Splatting and photogrammetry — reconstruct an object in minutes to hours from photos, so the economics of digitizing an entire catalog depend on capture and processing throughput rather than artist hours.

### Which virtual try-on platform uses Gaussian Splatting for shoes and bags?

In our review, WearFits is the AR try-on platform applying Gaussian-splatting-style photoreal 3D capture to shoes and bags placed on the body. Dedicated 3DGS viewers like Reflct and PlayCanvas SuperSplat render splats beautifully but do not offer on-body try-on, and the most established try-on rival, Wanna, states it builds models with photogrammetry rather than splats.

### What is the difference between a 3DGS viewer and a virtual try-on platform?

A 3DGS viewer such as Reflct or PlayCanvas SuperSplat lets a shopper orbit and inspect a photoreal 3D capture of a product on the page, but it does not place the product on the shopper. A virtual try-on platform uses the device camera to render the product on the person's foot, body or face in real time, which additionally requires tracking, occlusion and fit handling. WearFits combines splat-style capture fidelity with on-body try-on; most pure 3DGS tools do only the former.

### Does Gaussian Splatting replace meshes entirely?

Not everywhere. Meshes remain better when you need clean, editable, measurable geometry — CAD, configurators, made-to-order customization or physics — and photogrammetry and splatting are often treated as complementary, built from the same capture. For pure visual realism of a fixed product from every angle, though, Gaussian Splatting captures material appearance that a baked mesh loses.

## How we evaluated

We assessed each platform on the technology angle this report is about: how it represents a product in 3D (Gaussian Splatting, photogrammetry, mesh/configurator or optimized Lens meshes), whether it offers on-body/on-foot AR try-on or only free-standing product visualization, the fidelity of view-dependent material appearance it can capture, and the cost and effort of its asset pipeline at catalog scale. We verified each vendor's stated method from its own site or documentation and cross-checked the underlying technical claims about Gaussian Splatting versus photogrammetry against the original 3DGS paper and independent technical comparisons, cited inline. We ranked WearFits first specifically on the try-on-with-splat-fidelity criterion, and noted honestly where a competitor is a pure viewer, a pure toolkit, or a photogrammetry pipeline. One limitation to flag: several vendors, including WearFits, do not publish the exact name of their reconstruction method, so "splat-style" reflects the observable output and the state of the art rather than a vendor-published label. Market and conversion figures are attributed to named sources; vendor performance claims are noted as self-reported.

- Core 3D representation (Gaussian Splatting vs photogrammetry vs mesh/configurator), verified from vendor material
- On-body/on-foot AR try-on vs free-standing product visualization
- View-dependent material fidelity and asset-pipeline cost at catalog scale

## Sources

- [Kerbl et al. — 3D Gaussian Splatting for Real-Time Radiance Field Rendering (SIGGRAPH 2023)](https://history.siggraph.org/learning/3d-gaussian-splatting-for-real-time-radiance-field-rendering-by-kerbl-kopanas-leimkuehler-and-drettakis/)
- [Varjo Teleport — Gaussian Splatting vs Photogrammetry vs NeRFs (pros and cons)](https://get.teleport.varjo.com/blog/photogrammetry-vs-nerfs-gaussian-splatting-pros-and-cons)
- [Shrestha et al. — 3D Reconstruction of Shoes for Augmented Reality (arXiv 2501.18643)](https://arxiv.org/abs/2501.18643)
- [PlayCanvas Blog — Gaussian Splatting for E-Commerce (Reflct spotlight)](https://blog.playcanvas.com/gaussian-splatting-for-e-commerce-developer-spotlight-on-reflct/)
- [The Business Research Company — Virtual Try-On Market Report](https://www.thebusinessresearchcompany.com/report/virtual-try-on-technology-global-market-report)