FOSSDesk

THE OPEN-SOURCE OPTION / LIGHTROOM

Free, Open-Source
Lightroom Alternatives

Compare darktable and RawTherapee with documented RAW workflow tests, a two-file Lightroom sidecar migration check and official downloads.

Choose official downloads Read our test notes ↓

Hands-on tests, documented

darktable

Version
5.6.2
Released
2026-10-04
GitHub stars
13,437
License
GPL-3.0
Last checked
2026-11-12

RawTherapee

Version
5.13
Released
2026-07-26
GitHub stars
4,289
License
GPL-3.0
Last checked
2026-11-12
Official access

Direct from the source.

darktable

Windows
  • darktable-5.6.2-win64.exe ↗

    File version 5.6.2 · x86_64 · exe · 132.4 MiB

    Official SHA-2567f7a1eabcd0a1b548dcfacac842c8203fc85a4175697fb09f90bb891c8931edd
  • darktable-5.6.2-woa64.exe ↗

    File version 5.6.2 · arm64 · exe · 102.9 MiB

    Official SHA-256bcb700de8275f894c29ae4ccb5f600412b269bef1a581eecce9bbed697421fcc
macOS
  • darktable-5.6.2-arm64.dmg ↗

    File version 5.6.2 · arm64 · dmg · 84.8 MiB

    Official SHA-2566ff88e58a2a59cb07b0a1502fea7205e68cd783380a33a3ce7bda68ec29def0c
Linux

Checksums are supplied by the project; files have not been locally verified.

RawTherapee

Windows
macOS
Linux

Checksums are supplied by the project; files have not been locally verified.

Workflow

Tested scope and open questions.

Each result below is limited to the recorded samples and workflow.

  1. RAW import

    Three CR2, ARW and DNG samples opened in both GUIs. This is sample-specific evidence, not a claim about every camera or RAW format.

  2. Highlight recovery

    Controls and saved settings checked. Pending: recovery-quality assessment with suitable clipped RAW samples and reference exposures.

  3. Color adjustments

    Exposure and color controls exercised on one Canon CR2 in each GUI. No color-accuracy or comparative quality verdict.

  4. Photo organization

    Small-folder browsing/import checked. Pending: catalog management and large-library testing.

  5. Batch export

    Each GUI exported a three-image JPEG batch. Separately, CLI directory batches applied one whole saved profile to three RAW files; cross-camera dimension changes and a first-run repeat difference are recorded below. Pending: GUI batch copy/paste and large-job testing.

  6. Workflow migration

    Two Adobe-generated Lightroom Classic sidecars transferred four-star ratings, ordinary keywords, exposure and a Canon crop to darktable. Red labels became purple, and rendering and dimensions differed. Full Lightroom catalog, preset and RawTherapee migration remain untested.

Hands-on notes

What we actually tested.

Our newest migration check is below. Earlier November 12 experiments retain their original samples and limits.

Moving Lightroom edits: two genuine Adobe sidecars

Tested 2026-11-13 · Lightroom Classic 16.0 Windows GUI → darktable 5.6.2 · Two CC0 RAW copies · OpenCL disabled in the successful darktable run

We imported the Canon CR2 (983) and Sony ARW (2414) into a separate Lightroom Classic catalog, gave both four stars, a red label and the keywords CC0 and fossdesk-migration, and saved Adobe-generated XMP sidecars. The Canon received +0.75 EV and a crop; the Sony received −0.50 EV without a crop. A fresh darktable library read the sidecars; entering darkroom imported the adjustments. Both ratings and test keywords survived. The imported exposure instances read +0.750 and −0.500 EV, and the Canon had an enabled crop and rotate module. Six JPEGs covered Lightroom references, migrated darktable edits and darktable baselines without sidecars. Original RAW and Adobe-sidecar hashes stayed unchanged.

Scope and limits. This is partial sidecar migration, not a full Lightroom catalog import. Red did not survive correctly: Lightroom wrote the localized label “红色”, while both darktable labels became purple. The tested source used Chinese labels and the target used English; other language combinations were not tested. Rendering differed in color and brightness. Canon output was 4912×3181 in Lightroom versus 4929×3195 after migration; Sony was 6000×4000 versus 6048×4024. Identical module values do not mean identical processing. Collections, virtual copies, full history, presets, AI masks, GPS and hierarchical keywords were not tested; no Lightroom-to-RawTherapee transfer was tested. A first darktable process exited; the successful retry disabled OpenCL, without establishing the exit cause. A CLI backslash path returned 0 but wrote elsewhere; a forward-slash retry produced the expected files. These single-run exports are not a speed or quality benchmark.

Windows 11 x64, Intel Core i7-14700KF and approximately 31.8 GiB RAM. Official software packages were checked against their published SHA256 digests. Samples, separate settings, caches and outputs were kept on E:.

Three public RAW files from raw.pixls.us, each individually marked CC0: Canon EOS 5D Mark IV CR2 (983), Sony ILCE-7M3 ARW (2414) and Ricoh GR III DNG (3115). Source checksums matched and the RAW files remained unchanged. RAW sample archive ↗ · CC0 terms ↗

darktable: import, edit, save and export

Tested 2026-11-12 · darktable 5.6.2 · Windows GUI and command line · OpenCL disabled

The GUI imported three RAW files. On the Canon CR2, exposure changed from +0.700 to +1.000 EV, global vibrance to +10% and contrast to +5%. Highlight reconstruction was toggled and left on with inpaint opposed at a 0.950 clipping threshold. XMP edits saved and loaded; the settings and saved history persisted after restarting. One GUI export processed all three files to JPEG. Separate command-line runs produced JPEG and PNG for each sample.

Scope and limits. Pending: highlight recovery quality, Lightroom catalog/preset migration and large-library performance. The clipping-mask preview showed only a few small bright points; it does not establish recovery of broad overexposed areas. Only the CR2 received these edits; the other two files used default GUI processing.

RawTherapee: a saved profile and a three-image queue

Tested 2026-11-12 · RawTherapee 5.13 · Windows GUI and command line

The file browser recognized all three RAW files. On the Canon CR2, exposure was set to +0.50 EV, contrast to 5, saturation to 10 and highlight compression to 30, with highlight reconstruction enabled. A PP3 profile saved and reloaded after resetting exposure, restoring the edited values. The GUI queue exported three JPEGs with processing profiles. Separate command-line runs produced JPEG and PNG for each sample.

Scope and limits. Pending: highlight recovery quality, Lightroom edit/preset migration and large-library performance. The highlight controls were exercised, but recovery quality was not established. Only the CR2 received the listed edits. Initial automated numeric entry gave unintended values; button-based correction and PP3 checks confirmed the final settings. The console also emitted GLib warnings; successful exports do not mean a warning-free session.

All 12 command-line JPEG/PNG exports and six GUI JPEG exports decoded successfully and passed nonblank and dimension checks. The two final GUI CR2 JPEGs were also inspected visually; the other GUI exports received file checks only.

JPEG dimensions (darktable / RawTherapee): CR2 6744×4500 / 6732×4492; ARW 6048×4024 / 6016×4016; DNG 6000×4000 / 6012×4016. Different dimensions and processing defaults make this unsuitable for a matched-crop quality or speed ranking.

Batch parameters: one saved profile across three RAW files

Tested 2026-11-12 · darktable 5.6.2 and RawTherapee 5.13 · Windows CLI directory batches, not GUI copy/paste

A separate experiment applied the whole Canon CR2-authored XMP or PP3 to the same three CC0 samples. Each application processed the folder without an explicit profile, with the saved profile, and with that profile again. After correcting the output path, all 18 JPEGs decoded and passed nonblank checks. All six profile-versus-baseline pairs changed. RawTherapee’s six profile/repeat output PP3 files retained exposure +0.5, contrast 5, saturation 10, highlight compression 30 and enabled Coloropp recovery. For darktable, we checked the shared XMP input and changed outputs, not every module parameter in the exported files. RawTherapee dimensions stayed unchanged. Applying the whole XMP changed darktable’s Sony ARW from 6048×4024 to 5912×3980 and Ricoh DNG from 6000×4000 to 5976×4020; Canon CR2 stayed 6744×4500.

Scope and limits. Whole-profile application is not selective parameter transfer or a recommendation to reuse camera-specific settings across cameras. The cause of the dimension changes was not isolated to one module. In the first run, darktable returned exit code 0 but its backslash directory pattern wrote to an unintended relative folder; a separate forward-slash retry saved to the intended location. The first RawTherapee DNG repeat had a mean absolute channel difference of about 0.00010–0.00014 on a 0–255 scale. All six repeat pairs in the retry were pixel-identical, but that does not erase the first-run difference or establish deterministic output; its cause remains unknown. Pending: GUI batch copy/paste, selective transfer, cross-camera color correctness, highlight recovery quality, Lightroom migration and large-job reliability. Full visual-quality assessment was not performed. The earlier GUI exercise edited only the CR2; this is a separate CLI experiment.

Pending: a full comparison.

Highlight recovery quality, full Lightroom catalog/preset migration, Lightroom-to-RawTherapee transfer, GUI batch copy/paste and large-library behavior remain unverified. These tests support a small import/edit/export workflow, not a recommendation for every photographer.

Getting started

Choose your platform.

Official documentation checked 2026-11-12 · Not an installation test

darktable publishes Windows installers, macOS disk images and Linux AppImages. Its installation page currently states macOS 13.5 or newer and no official Intel Mac releases. Check the package architecture before downloading. darktable installation guidance ↗

RawTherapee directs users to stable releases on its website or a package manager. Its documentation warns that development builds are not intended for production use and may change sidecar-file results. RawTherapee stable and development builds ↗

Observed Windows setup: the official darktable 5.6.2 installer completed a per-user installation to E: without a restart; the official RawTherapee 5.13 ZIP was extracted there. Both GUIs launched. This is a record of those packages, not verification of every installer or platform.

Common questions

Before you switch.

Which option fits my workflow?

Both tools completed a small Windows import, edit, save and export exercise. This does not establish which fits your workflow better. In that GUI exercise, only one Canon CR2 received the recorded adjustments; highlight recovery quality and large-library performance remain unverified. A separate two-file Adobe-sidecar migration check is described below; full catalog and preset migration remain untested.

Will my existing edits transfer?

Some sidecar data transferred in our two-file Lightroom Classic 16.0 to darktable 5.6.2 check: four-star ratings, two ordinary keywords, exposure values and a Canon crop. Red labels became purple, and colors, brightness and output dimensions differed. This was a genuine Adobe XMP transfer into a fresh darktable library, not a full .lrcat catalog import. Collections, virtual copies, full history, presets and AI masks remain untested. No Lightroom-to-RawTherapee migration was tested. Keep your original catalog and files, and evaluate separate copies before switching.

Which RAW files are supported?

Our three samples were Canon EOS 5D Mark IV CR2, Sony ILCE-7M3 ARW and Ricoh GR III DNG. Both applications opened these files and exported JPEGs in the GUI, plus JPEG and PNG in separate command-line tests. This is evidence for these files and tested versions, not a complete camera-support list.

How do the tools handle large libraries?

Pending: large libraries were not tested. Each GUI batch contained three images. No measured catalog-scale performance, large-job reliability or comparative speed conclusion is available.

What should I check before switching?

The remaining checks are your own RAW files, saved edits, output dimensions and any existing Lightroom catalog or presets. Our exports differed in dimensions between applications, and the experiment did not establish matched-crop quality, highlight recovery quality or full catalog and preset compatibility. The separate two-file Adobe-sidecar check retained some metadata and exposure values, with label and rendering differences. A separate CLI directory test applied one whole saved PP3 or XMP across three RAW files; this was not GUI batch copy/paste. All 18 retry JPEGs decoded. Applying the CR2-authored XMP changed darktable’s ARW from 6048×4024 to 5912×3980 and DNG from 6000×4000 to 5976×4020. RawTherapee’s six output PP3 files retained the checked settings, but its first-run DNG repeat had a small pixel difference. Retry pairs matched; this does not establish deterministic output or cross-camera correctness. GUI parameter transfer, selective copying and large-job reliability remain unverified. Retain your originals and existing edits while evaluating a separate test copy.