If the image is a Bayer RAW file, raw chromatic aberrations is one candidate TCA path; if you use it, disable TCA in lens correction. Supported embedded metadata or a correctly matched Lensfun profile can provide another candidate path. Compare candidates one at a time on the same high-contrast edge; neither data source is documented as inherently more reliable. Do not enable RAW CA and lens-correction TCA together for the same defect: darktable 5.6 warns that they can conflict and produce over-correction artifacts.
Make the decision in order
- Confirm the input type. Check the image information and the camera file type.
raw chromatic aberrationsis defined for RAW images recorded with a Bayer sensor. It does not apply to other sensor types, and it does not correct images identified as monochrome. The visible color of a fringe is only a clue; use the actual input type and image metadata as the decision evidence.
- Inspect a high-contrast edge. Use a branch, roof wire, backlit hair, or another dark-to-bright boundary near the frame edge. Compare the uncorrected view at a useful zoom. This is an editorial comparison method, not a darktable diagnostic guarantee. Look for colored seams and for changes in sharpness after correction.
- Check the lens path. In
lens correction, see whether the camera and lens are correctly identified and whether the selected method isLensfun databaseorembedded metadata. Lensfun uses camera, lens, focal length, aperture, and sometimes focal distance. The module also reports which corrections were actually applied. If the camera documentation or supported metadata explicitly confirms that an in-camera correction has already been made, remove only that duplicate fromcorrections; otherwise record the in-camera status as unknown. A missing or incorrect profile is a reason to question the automatic TCA result, not to stack another TCA correction automatically.
- Select one primary route. For a supported Bayer RAW file, compare the available TCA sources by enabling only one route at a time: either
raw chromatic aberrationswith TCA disabled inlens correction, or TCA from supported embedded metadata or a correctly matched Lensfun profile withraw chromatic aberrationsdisabled. The manual does not establish a priority rule when a profile is missing or when an in-camera correction is undocumented. The module expects good white-balance data; its default settings are often sufficient according to the manual, but the result still needs visual review.
- Compare the seam and nearby detail. Review the same high-contrast edge before and after the chosen route. Stop when the colored seam is reduced without creating a new colored rim, tint, or loss of edge detail. If the result is uncertain, compare two versions with only one route enabled at a time. This isolates the choice; it is an author-recommended comparison procedure, not a documented automatic test.
The lens module can handle more than TCA: its correction choices can include distortion, TCA, and vignetting. If the camera has already performed part of that work, change the corrections selection so that only the needed corrections are applied. That setting does not justify enabling a second TCA algorithm.
Acceptance worksheet
Use this worksheet for one image. The observations and thresholds below are hypothetical review aids, not values promised by the manual.
| Check | Hypothetical observation | Decision |
|---|---|---|
| RAW sensor label | RAW file is identified as Bayer; not monochrome | raw chromatic aberrations is eligible |
| Lens metadata | Camera and lens are recognized, but the profile reports no usable TCA | Record this profile TCA path as unavailable; assess only another eligible path. Missing profile data does not establish that another route is better |
| High-contrast edge | A cyan-red seam is visible on a dark roof wire near the right edge | Inspect that same wire after each isolated comparison |
| RAW route comparison | Seam is smaller, while nearby roof detail remains visually stable | Keep RAW route as the single TCA path |
| Embedded or Lensfun route | A correctly matched profile supplies TCA and produces the cleaner comparison | Use lens TCA instead; disable RAW CA |
| Internal camera correction | Camera documentation or supported metadata confirms vignetting correction, but TCA status is unknown | Disable the confirmed duplicate vignetting correction; record TCA as unknown and do not infer it |
| Non-Bayer input | File is not Bayer RAW, or is identified as monochrome | Do not use RAW CA; consider the RGB module path |
The table does not infer a hidden correction from a color count, pixel size, or fringe width. It records an observed comparison and the route chosen for that image.
Limits and next steps
raw chromatic aberrations is limited to Bayer RAW input. For non-Bayer or already-rendered RGB data, the chromatic aberrations module is an available later RGB-stage path; it does not require RAW input and is not the same operation as correcting Bayer data before demosaic. The documented warning about conflicting corrections is specifically between RAW CA and TCA in lens correction; it does not require disabling unrelated lens corrections such as distortion or vignetting. When evaluating RGB chromatic aberrations, compare the same edge and nearby color and detail: the module workflow advises reducing strength to avoid washing out colors. A lens-correction TCA path may still be available for an image where RAW CA is inapplicable, provided supported correction data exists. The choice to compare paths one at a time is an author-recommended isolation procedure, not an official automatic test.
If lens identification fails, first inspect the camera and lens fields and the available correction method. An adapter, third-party lens, missing Lensfun profile, or stale metadata can prevent a useful automatic match. The documentation describes manual selection and Lensfun data maintenance as possible remedies, but it does not guarantee that every lens has a profile. Do not present a guessed profile as calibration.
If lens correction changes the borders, inspect the crop and scaling result as well. Strong corrections can leave visible repeated border pixels. That is a separate border consequence of geometric correction, not evidence that TCA should be applied again.
Three quick questions
Does a purple or green edge prove that RAW CA is the right choice?
No. Color is only a visual clue. Confirm Bayer RAW eligibility and compare the available correction path on the same edge.
Can I use lens correction for distortion and RAW CA for color?
The documented conflict concerns TCA specifically. You may need lens correction for other corrections, but disable its TCA component when using RAW CA. Choose the controls separately and verify what the module reports as applied.
What if neither route removes the fringe?
For a non-Bayer or RGB image, try chromatic aberrations with a restrained comparison. For a supported RAW file, recheck white-balance data, profile identification, and whether the visible edge is actually caused by another processing step. The documentation does not promise recovery of detail that was never recorded.
Sources: lens correction, raw chromatic aberrations, and chromatic aberrations.
Sources and next step
- darktable 5.6 user manual - lens correction
- darktable 5.6 user manual - raw chromatic aberrations
- darktable 5.6 user manual - chromatic aberrations
This guide interprets public documentation and uses explicitly hypothetical acceptance examples, not application tests. Controls and availability can change; verify the cited manual against your installed version.
For work created in Imagild, carry the input requirements and acceptance checklist into the photo editor. The external software controls discussed above belong to their respective products.