X-Glass
BrowseCompareAboutGet App
BrowseCompare

About X-Glass

01Background02Coverage & Roadmap03Data & Accuracy04Disclaimer05Privacy06Support the Project07Acknowledgments

Background

The Fujifilm lens ecosystem — spanning both the X Mount (APS-C) and G Mount (medium format) systems — has grown from a handful of native options in its early days to today's landscape of dozens of brands and hundreds of lenses. Yet specs are scattered across brand websites and various platforms in inconsistent formats, making meaningful comparison a slow and frustrating exercise. X-Glass was built to pull all of that fragmented information into one place — a normalized, comparable database that lets the data speak for itself.

I'm a web front-end engineer, and X-Glass is a project I develop and maintain entirely on my own. I've always had a strong conviction and passion for building products that don't compromise on engineering quality, user experience, or aesthetic design — and that drive is one of the key forces behind why this project exists.

Coverage & Roadmap

X Mount

BrandActiveDiscontinuedLenses
Fujifilm✓✓49
Sigma✓○11
Tamron✓○4
Viltrox✓—14
7Artisans✓—27
TTArtisan✓—21
Brightin Star✓—15
SG-Image✓—11
Laowa○——
Meike○——
Sirui○——
Voigtländer○——
Total152

✓ Covered · ○ Under evaluation · — No plans

G Mount

BrandActiveDiscontinuedLenses
Fujifilm✓—19
Total19

✓ Covered · ○ Under evaluation · — No plans

Missing a lens you're looking for?

Data & Accuracy

All data in X-Glass is produced by two independent pipelines. The spec pipeline collects and structures lens specifications from official brand websites. The pricing pipeline independently samples market prices from e-commerce storefronts. The two are fully isolated, each with its own data sources and quality control process.

All data defers to its official source. Spot something off?

Spec data

Data comes from official brand websites only

Collection only accesses brand official sites; no third-party material is used.

Deterministic fields are computed in code

Focal length, aperture, series name, and other fields strictly derivable from the lens ID are all computed in code. Fields requiring semantic understanding are extracted by the model from the collected official text — every value is traceable back to its source.

Strict stage isolation

Pipeline stages execute in a fixed order. Each step uses only the previous step's output; intermediate stages cannot consult external sources or revise prior conclusions.

Independent cross-check before publish

A separate AI without access to the pipeline's context samples and compares against published reviews before publish, to surface errors that may exist at the source itself. These third-party materials are used only for verification, not as data sources.

Spec data is produced through a multi-stage pipeline:

  1. 0
    DiscoveryScripts index lens listings from each brand's official website
  2. 1
    Raw Data RecallAI Agents collect official lens descriptions and raw spec text from manufacturer sources
  3. 2
    StructuringScript rules compute deterministic fields; AI Agents derive semantic fields
  4. R
    Human ReviewMaintainer reviews each entry and applies necessary corrections
  5. P
    Publish GateSchema validation, normalization, independent cross-check, and final merge

The database is updated when new lenses launch or when existing data needs correction. The version number and last-updated timestamp in the footer reflect the latest published data release.

Price data

Produced independently from spec data

Pricing has its own sampling sources and collection process, kept separate from the spec pipeline.

Sourced from brand-official storefronts

Prices come from brand-operated e-commerce. When a brand has no unified global storefront, the US storefront is used as a reference.

Used market as fallback

When no official listing is available, used market listings are sampled, and a median of multiple listings is computed to reduce outlier impact.

Point-in-time snapshots, not real-time

Prices reflect the state at sampling time and are not continuously updated.

Full pipeline architecture

Detailed Mermaid diagram of data flow, branching, and human/AI nodes.

View on GitHub

Disclaimer

X-Glass data is sourced from public materials and refined through manual review. We make every effort to keep specs accurate and up to date, but errors and delays are inevitable. Always verify key specifications on the manufacturer's official page before making a purchase decision.

X-Glass is not responsible for any decisions made based on information presented here.

X-Glass is an independent third-party tool and is not affiliated with, endorsed by, or sponsored by Fujifilm or any lens manufacturer.

Product images and brand trademarks are the property of their respective owners. If you are a rights holder and believe your content has been used improperly, please contact xglass@sentacraft.com.

Privacy

X-Glass does not require registration, does not collect personal information, and does not use cookies.

This site uses Vercel Analytics to collect aggregated, anonymous data such as page views, referrers, and device types — solely to help improve the product. No personally identifiable information is collected.

Support the Project

X-Glass is open source on GitHub. Every spec in this database went through tedious data extraction and manual cross-checking. If it helped you find the right lens — or saved you some time — consider supporting the project.

Support on Ko-fi

Acknowledgments

Over the past year, large language models have shown capabilities that once felt out of reach. I could not have imagined completing a project at this scale on my own — let alone shipping it in under a month. Without large language models and AI coding agents, this project might have stayed an idea forever. I am grateful to live in an era where the technology dividend is real, and where a spark of inspiration can actually become something in the hands of a solo developer.

Iris is X-Glass's resident character — the aperture animation on the home page that opens and closes with each tap.

Building Iris required Claude to first study the mechanical principles behind petal-shaped aperture blades — working through structural diagrams, motion logic, and kinematics documentation before writing a single line of code. What it produced is a genuine three-body constraint kinematic engine: forward and inverse kinematics solved analytically in real time, aperture openings computed from arc-intersection geometry. I had no background in mechanism kinematics whatsoever. The entire system was built from scratch.

Iris is more than an animation. Like Clippy once lived in the corner of Microsoft Office, Iris is X-Glass's resident character — accompanying every visitor with a quiet personality of its own.

Gemini shaped several key design decisions throughout the project. When building the Filter Panel, it established the interaction language for single- and multi-select controls so users could understand a control's behavior at first glance. On the lens list, it refined card layout and spacing across screen sizes, keeping the information hierarchy clear on every device. For the share poster, it pushed the design toward something genuinely clean, restrained, and visually considered.

As an engineer, I could feel when something in the design was off — but rarely knew why. Gemini gave me the vocabulary to diagnose design problems, and the understanding to know why the fix worked. The site reached a level of craft I couldn't have gotten to on my own.

My deepest gratitude to the AI collaborators whose contributions were decisive in making this project a reality.