OpenRevelare

An open-source digital darkroom for film negatives
Turn camera-scanned RAW and scanner TIFF files into positives, one roll at a time.

English · 简体中文

Download · First roll · Overview · Interface · Features · Documentation · Build · Licence

---

Overview

New users can start with Your first roll, also available offline in Help → documentation → First roll.

OpenRevelare is a standalone desktop application for photographers who digitise film with a camera or scanner. Import a roll, share its base calibration across frames, make individual edits and export your photographs. No subscription or host photo editor is required.

Colour negatives contain an orange base and dye layers. OpenRevelare linearises the input, removes the base, white-balances and inverts it in the Cineon log-density domain, then exports a positive.

• Explainable: base, maximum density and channel-balance parameters have defined meanings.
• Reproducible: settings are stored in a .ncproj beside the image; source files remain untouched.
• Local-first: image processing works offline without an account. Update checks and optional model downloads use the network. The inversion pipeline runs in the shared CPU Core; supported Windows/macOS systems can accelerate selected resizing, sharpening and LUT operations on the GPU.
• Cross-platform: C# / .NET 8 + Avalonia for Windows, Linux and macOS.

The application is intended for whole-roll camera-copy and scanning workflows, archival parameters and reproducible reprocessing. HDR editing and export are supported; Linux currently keeps the on-screen preview on an explicitly labelled SDR fallback. It does not provide per-roll colour-chart calibration; use DiVERE for heritage, commercial-archive or research work that requires it.

Installation

Packages are published on Releases with the .NET runtime included.

Platform	Package	Status and requirements
Windows 10/11 x64	setup.exe	Stable; no additional dependency
Linux x86_64	.AppImage	Beta; glibc ≥ 2.35
macOS Apple Silicon	-arm64.dmg	Beta; macOS 12+, no real-device validation yet
macOS Intel	-x86_64.dmg	Beta; macOS 12+, no real-device validation yet

Platform notes

• Windows: if SmartScreen appears, select More info → Run anyway.

• macOS: approve the first launch under System Settings → Privacy & Security → Open Anyway, or run:

  xattr -dr com.apple.quarantine /Applications/OpenRevelare.app

• Linux:

  chmod +x OpenRevelare-*.AppImage && ./OpenRevelare-*.AppImage

  Add --appimage-extract-and-run if the package cannot start directly. Linux currently previews through SDR; HDR editing and export remain available.

Model

Stage	Scope	Responsibility
FilmBase	Whole roll	Base transmittance, maximum density, channel balance and inversion parameters
SceneBase	Single frame	Temperature, exposure, contrast, saturation and curves

The separation allows physical parameters to be shared across a roll while frame-level edits remain independent.

Principles

1. The mask is physical: measure the base; do not replace measurement with visual guessing.
2. Density is the computation domain: the mask is approximately a constant offset, making white balance and inversion predictable.
3. Restoration and creation are separate: FilmBase provides consistency; SceneBase provides expression.

Workflow

1. Import RAW, scanner TIFF or another supported image and enter roll metadata.
2. Calibrate the roll: estimate the base, white balance, shadow endpoint and d_max; correct manually when required.
3. Apply FilmBase to the roll or selected frames.
4. Edit frames in SceneBase with temperature, exposure, contrast, saturation and curves.
5. Export TIFF, JPEG, HEIC, linear DNG or scene-linear ACEScg TIFF. Start with an sRGB JPEG for sharing, or a 16-bit TIFF for further editing.

Changes are written automatically. The .ncproj is placed beside the source when that directory is writable; read-only media use the per-user rolls fallback shown below.

Interface and examples

Cineon calibration

_{Roll thumbnails, the current frame and shared physical parameters on one page.}

Display adjustment

_{Frame-level creative adjustments without changing FilmBase.}

Roll management, filtering and project recovery	Full-roll contact sheet with sprocket layout and metadata

Reversal result

Raw input	NLP automatic reversal	OpenRevelare result

Features

Category	Capabilities
Reversal	Cineon density-domain inversion, auto-calibration, narrow-band light decoupling (Path A), black-and-white negatives
Colour	ACEScg workspace; sRGB, Display P3 and Adobe RGB output; exact ICC management; Stage 2 controls
Pre-processing	LCC flat-field, offline Lensfun lens matching, distortion and vignetting correction, sprocket masking and geometry crop in linear light
Auto crop	135 positive / negative auto crop: content 3:2 and 35 mm border 37:35 modes, small-angle straightening, one crop scale and pixel ratio across the roll
Workflow	Roll management, roll sync, virtual copies, format presets, 80-step undo/redo, histogram and waveform
Output	16-bit TIFF, JPEG, 10-bit HEIC, linear DNG, 32-bit float scene-linear ACEScg TIFF and HDR gain-map JPEG; templates, scaling and sharpening
Look / HDR	Cineon-input .cube print looks, SDR/HDR roll rendering, HDR limit in stops, HDR contact sheets; direct PQ or Adaptive HDR HEIC (viewer support required)
Reports	Per-frame report of input domain, base source and inversion parameters

Inputs include DNG, NEF, CR2/CR3, ARW, RAF, RW2, ORF, PEF, IIQ, Hasselblad/Flextight .fff and TIFF. JPEG/PNG are export or cache formats, not negative-input formats. RAW file-dialog filters accept upper- and lower-case extensions.

How it works

Transmittance T is converted to log density:

$$D = -\log_{10}(T)$$

The core operations are:

$$T_{norm} = T / T_{base}$$

$$D_{corr}[c] = D[c] \times w_{high}[c] + w_{offset}[c]$$

$$T_{pos} = 10^{\left(\frac{R_{out}}{D_{max}[c]}D[c]-R_{out}\right)}$$

The per-frame pipeline is: linearisation → optical correction → optional light-source decoupling → base removal → density-domain white balance → inversion → output. The application includes the complete derivation under Help → Operation guide / Technical theory.

Data locations

Content	Windows	Linux / macOS
Settings and roll index	%APPDATA%\OpenRevelare	$XDG_CONFIG_HOME/OpenRevelare/ (default ~/.config)
Contact-sheet cache	%LOCALAPPDATA%\OpenRevelare\sheets	$XDG_CACHE_HOME/OpenRevelare/sheets/ (default ~/.cache)
DNG decode cache	.revelare-cache/ beside the source	Same
Print LUT drop-in folder	%LOCALAPPDATA%\OpenRevelare\luts	$XDG_DATA_HOME/OpenRevelare/luts/ (default ~/.local/share)
Project file	.ncproj beside the source when writable; otherwise %APPDATA%\OpenRevelare\rolls	Beside the source when writable; otherwise $XDG_CONFIG_HOME/OpenRevelare/rolls/

Cache locations and limits are configurable. Uninstalling does not remove these data.

Documentation

• Your first roll: import, calibration, editing, export and reopening a project.
• Operation guide: controls and everyday workflows.
• Technical theory: the colour and inversion model.
• Changelog: release history and improvements.
• Contributing · Security policy · Third-party notices.

Frequently asked questions

Can I import JPEG or PNG negatives? No. Use a supported RAW file or TIFF; 16-bit TIFF is preferable to 8-bit for shadow detail.

Do edits change my originals? No. Edits are stored in .ncproj project files. For a complete backup, retain the originals, projects, LUTs and calibration references together; an exported TIFF does not replace them.

Why does HDR look different in another app? HDR display depends on the operating system, display and viewer. Linux uses an SDR preview fallback. Check a sample export in your intended viewer before exporting the full roll.

Is Smart Cast Correction required? No. Deep-WB is optional and downloaded separately after licence acceptance. Its CC BY-NC-SA 4.0 non-commercial licence differs from the application's GPL licence; review model licensing before use.

Building from source

Requires .NET 8 SDK, Python 3.12+, CMake 3.21+ and a C/C++ toolchain (Visual Studio 2022 C++ on Windows, Xcode command-line tools on macOS, GCC/Clang on Linux). x64 HEIC builds also require NASM 2.13+ on PATH. The first native build downloads checksum-pinned dependencies.

Clone the repository:

git clone https://github.com/Toshihiko-Lin/Open-Revelare.git
cd Open-Revelare

Build native dependencies on Windows (PowerShell):

./packaging/lcms/build-windows.ps1
./packaging/libraw/build-windows.ps1
python packaging/compute/build.py
python packaging/heif/build.py
./packaging/windows/build-win32-presenter.ps1

On Linux / macOS:

bash packaging/lcms/build-unix.sh
bash packaging/libraw/build-unix.sh
python3 packaging/compute/build.py
python3 packaging/heif/build.py

Then build, test and run:

dotnet build -c Release -warnaserror
dotnet test -c Release --no-build
dotnet run --project src/OpenRevelare.Gui -c Release

LibRaw · Little CMS · HEIC · GPU compute. See RELEASE.md for platform packaging and release checks.

CLI example:

dotnet run --project src/OpenRevelare.Cli -- -i neg.tiff -o pos.tiff --input-linear --color-space sRGB

--input-linear and --input-srgb are mutually exclusive input declarations. Legacy --d-max, --grade, --pivot and --scan-exposure-ev are accepted for old parity scripts but are no-ops; the current inversion is controlled by the two calibrated density endpoints and output range.

Read CONTRIBUTING.md before development. Calibration experiments are documented in docs/calibration/; colour-management decisions are recorded in docs/color-management-architecture.md.

Licence and third-party components

The code is licensed under GPL-3.0-only; see LICENSE.

Deep-WB is an optional download: accept the license when first using Smart Cast Correction, or run OpenRevelare.Cli --install-deep-wb. Default packages do not include model weights. Downloads automatically fall back to the Gitee mirror if GitHub fails; offline installation is also supported. The ONNX runtime remains bundled.

models/net_awb.onnx is the Deep White-Balance Editing model, distributed under CC BY-NC-SA 4.0 and outside this project's GPL scope. See models/README.md and THIRD_PARTY_NOTICES.txt.

References include LightSourceDecouple, DiVERE and darktable negadoctor.

Project origin and acknowledgements

The project began as a personal film-copying tool. After user feedback and a core rewrite, it was released as a cross-platform open-source project in August 2026. Early development was supported by:

• 豆腐
• Caramello_焦糖玛奇朵
• REPEATER000
• jamais
• hhd

Support the project

The project is maintained independently and released under GPL-3.0-only. If it is useful to your workflow, support is welcome:

Limitations and feedback

• No per-roll colour-chart calibration.
• 8-bit TIFF may show banding in shadows; 16-bit input is recommended.
• Linux uses an SDR preview fallback; HDR editing and export remain available.
• macOS native presentation still requires real-device validation; TIFF/CPU rendering remains available.

Report issues through GitHub Issues with the OS version, camera or scanner, input format and error details. Do not upload private originals.