README.md
September 1, 2026 · View on GitHub
An open-source design, analysis, and optimization environment for optical thin-film coatings.
Website · Tutorials · Live demo · Documentation · Download · Roadmap
English · 简体中文

What is TFStudio?
TFStudio is a desktop application for designing and analyzing optical thin-film coatings: antireflection coatings, mirrors, beamsplitters, bandpass and edge filters, and more. It provides a double-precision optical engine, refinement and synthesis algorithms, and an analysis suite, in a docked, multi-window interface.
⚠️ Status: TFStudio is independently developed software. Always verify critical designs against your own calculations and measurements before committing them to a production deposition run.
Key features
Design & evaluation
- Transfer-matrix method (TMM) for absorbing and dispersive media at oblique incidence, both s- and p-polarization
- Full-system modeling: front coating, substrate (with absorption), and back coating, including incoherent substrate multiple reflections
- Reflectance / transmittance / absorptance spectra, color, integral figures of merit
- Layer editor with simultaneous physical / optical / quarter-wave / full-wave thickness representations
Optimization & synthesis
- Damped least-squares / Levenberg-Marquardt refinement with an analytic Jacobian
- Additional refiners: Newton, Newton-CG, SQP, conjugate-gradient, differential evolution, simulated annealing
- Needle optimization and gradual evolution synthesis (automatic layer insertion from scratch)
- Structural optimization over the layer count itself
- Flexible merit function: spectral targets, ramps, band averages, worst-case operands, thickness constraints
- Fitting to an imported measured spectrum, as one more row in the merit function
- Multi-threaded via a Web Worker pool; hot kernels accelerated with WebAssembly
Analysis windows
- Optical evaluation, admittance diagrams, electric-field profiles, group delay / GDD, ellipsometric parameters, color evaluation, refractive-index profile, layer thickness diagram
- Tolerance & manufacturing analysis: Monte-Carlo error analysis, layer sensitivity, inhomogeneity, roughness/scattering, systematic deviations
Materials
- Built-in library generated from the refractiveindex.info database (CC0)
- Sellmeier / Cauchy / tabulated dispersion; complex index with explicit conventions
- Import of external catalogs and an in-app refractiveindex.info browser
Measured data
- Import of measured R / T / A spectra and of ellipsometric Ψ and Δ, drawn over the calculated curves
- n,k characterization: derive a film's index, extinction and thickness from a measured transmittance and reflectance, or from a Ψ and Δ pair, and save the result as a material
- Export of measured or calculated curves in nm, µm or wavenumber, as a fraction or a percentage
Manufacturing
- Deposition / monitoring simulation (broadband and monochromatic optical monitoring)
- Monitoring worksheet: per-layer monitoring wavelength and witness-chip assignment, with the layers that cannot be terminated closely enough flagged before the run
- Process exporter and optical-coating data interchange (including Zemax OpticStudio coating export/import)
Platform
- Cross-platform desktop app (Electron + React, pure JavaScript)
- Tabbed ribbon with a search box that finds any tool by name
- Windows dock, or tear off the layout onto a second monitor
- Built-in help/documentation, English, Russian and Chinese UI
Scientific basis
Methods and their sources:
- Transfer-matrix method: H. A. Macleod, Thin-Film Optical Filters, 5th ed.
- Numerical needle synthesis: Sullivan & Dobrowolski, Appl. Opt. 35, 5484 (1996); Tikhonravov et al., Appl. Opt. 35, 5493 (1996)
- Gradual evolution: Tikhonravov et al. (2007)
All computations are double precision.
The transfer-matrix engine is published separately as tmmcore. Its comparison page measures it against tmm (Byrnes), tmm_fast, tmmax and tmm_faster on accuracy and speed, and states the method and its caveats. Reference outputs are committed, so npm run compare reproduces the accuracy table with Node alone and no Python.
Installation
Download (recommended)
Grab the latest build for your platform from the Releases page.
Windows: TFStudio Setup <ver>.exe installs normally; TFStudio-<ver>-Portable.exe is a single executable that needs no installation, for locked-down deposition PCs. Separate Windows 7/8.1 builds are published alongside.
Linux: On Debian and Ubuntu, TFStudio-<ver>-amd64.deb is the recommended package:
sudo apt install ./TFStudio-*-amd64.deb
tfstudio
Installing as root is what lets the Chromium sandbox stay enabled. The .deb is the only Linux package that keeps it on, and it also adds TFStudio to the applications menu.
TFStudio-<ver>-x86_64.AppImage is the portable alternative:
chmod +x TFStudio-*-x86_64.AppImage
./TFStudio-*-x86_64.AppImage
AppImages need FUSE 2, which Ubuntu 22.04 and later no longer install by default. Either add it (sudo apt install libfuse2), run the AppImage with --appimage-extract-and-run, or use the TFStudio-<ver>-x64.tar.gz archive, which unpacks and runs with no such dependency.
From the .tar.gz, start the app through the bundled launcher rather than the binary:
tar xzf TFStudio-*-x64.tar.gz
cd TFStudio-*-x64
./tfstudio.sh
tfstudio.sh works around a Chromium requirement that an unpacked archive cannot satisfy: the sandbox helper has to be owned by root, which an archive extracted as a normal user never is. Running the tfstudio binary directly instead aborts on startup with The SUID sandbox helper binary was found, but is not configured correctly. The launcher disables the sandbox only when there is no working alternative, so installing the .deb remains the more secure option.
Want to try it first? Run the live web demo for example designs and live spectra in the browser, with no installation required.
Build from source
Requires Node.js 18+ and git.
git clone https://github.com/aai2k/TFStudio.git
cd TFStudio
npm install
npm start # launch the app
The WebAssembly transfer-matrix kernel arrives prebuilt with the tmmcore
dependency, so no Emscripten toolchain is needed and source builds get the same
performance as the released binaries.
npm run build checks out the refractiveindex.info database submodule and installs
the docs-site dependencies automatically. The database is large; to pull it upfront
instead of on first build, clone with --recursive.
Other useful scripts:
npm test # run the test suite
npm run docs:install # install docs-site deps (needed before docs:dev)
npm run docs:dev # preview the documentation site
npm run build # package a distributable (electron-builder)
Packaging releases
build-release.ps1 provisions everything a fresh clone needs and packages the
installers in one step. It asks which optional targets to include; pass the flags
to answer up front.
npm run dist # Windows 10/11 installer + portable
npm run dist -- -Win7 # ...and the Windows 7/8.1 builds
npm run dist -- -Linux # Linux .deb + AppImage + tar.gz only
The Linux artifacts are produced by build-release-linux.sh, which the release
script drives through WSL; it also runs on any Linux host directly. It needs a
distribution with Node.js 18+ and rsync, and builds in the Linux filesystem
rather than in place, so a Windows checkout keeps its Windows node_modules.
The Windows-driven WSL build skips GUI verification. A direct Linux build launches
the unpacked application under Xvfb as a smoke test, or skips it with a message if
Xvfb is not installed.
Note that the smoke test runs under Xvfb, so it exercises the X11 path only. A regression that appears solely under native Wayland, such as a window that is never presented, will pass it.
macOS builds require a macOS host and are not currently published.
User documentation is hosted at docs.tfstudio.xyz, is built into the app (Help menu), and its source lives in docs-site/.
Citing TFStudio
If TFStudio contributes to your work, please cite it. Citation metadata is in CITATION.cff; GitHub renders a "Cite this repository" button from it.
Contributing
Issues and pull requests are welcome. Because TFStudio is a scientific tool, contributions to the optical engine are held to a physics-correctness bar (cite your sources, validate against a reference, add a test). Please read CONTRIBUTING.md before opening a pull request.
By contributing you agree that your contributions are licensed under the project's MIT license.
License
MIT © 2026 Andrey Achapovsky
Author
Andrey Achapovsky: ORCID 0009-0005-1497-6279
Acknowledgements
- Material data derived from the refractiveindex.info database (CC0, public domain).
- Built with Electron, React, Apache ECharts, and KaTeX.