A 3D LUT can clip highlights or shadows in darktable when image values fall outside the LUT module’s [0,1] range: the module clips those values, and the manual notes you may need to reduce the input range. Before changing settings, confirm whether the LUT is a grading/look LUT or a camera-log conversion LUT, and identify its intended application color space. Changing interpolation is a separate check; it does not expand the input range or restore clipped detail.
Diagnose the LUT and its input
Use these steps to narrow down the cause on a working copy of the image:
- Identify the LUT from its documentation or reliable metadata, not its filename or thumbnail. A grading or film-look LUT is intended for a neutral image and, by default, the module sits after filmic rgb. A camera-log conversion LUT has a different purpose and placement; the manual says that use case has not yet been tested in darktable.
- Confirm the module’s application color space matches the space for which the selected LUT was built. The manual notes that
.cubefiles are usually associated with Rec. 709 and many other LUTs with sRGB, but these are clues, not substitutes for the LUT’s actual type or documentation. - Observe the image immediately before and after applying the LUT. Note whether bright or dark regions become flat or lose visible variation. This comparison is an author-recommended diagnostic, not a documented clipping indicator or a software test.
- If the LUT is a grading LUT and clipping is plausible, reduce the range of its input using an appropriate preceding adjustment, then compare again. Record the adjustment rather than assuming it will improve every LUT or image.
- As a separate check, change the interpolation method and compare. Keep this result distinct from the input-range comparison.
Record a useful acceptance check
The table below is a hypothetical observation log, not a predicted darktable result. Its before-and-after notes illustrate how to decide what to investigate next; they are not measurements or guarantees. “Linear or display-referred” describes the input type categories documented for the module, not a value to infer from how the picture looks.
| Check | Hypothetical observation | Decision |
|---|---|---|
| LUT purpose | Documentation identifies a creative grading LUT, not a camera-log conversion | Follow the grading workflow; do not reposition it as a log conversion on this evidence |
| Input and color space | Hypothetically, the LUT documentation specifies Rec. 709 while the module’s application color space is set to sRGB; the input is separately recorded as display-referred | Treat this as a suspected mismatch, not as evidence that Rec. 709 follows from display-referred input. First verify the LUT’s actual authoring color space; only if Rec. 709 is confirmed, select Rec. 709, then reassess the input range |
| Bright regions | Before the LUT, a bright cloud has visible tonal variation; after it, the region appears flat | Record suspected highlight clipping; this visual cue alone does not prove values exceeded [0,1] |
| Dark regions | Before the LUT, a dark coat shows folds; after it, folds are less distinct | Record suspected shadow clipping separately from the bright-region observation |
| Range comparison | After a preceding adjustment reduces the input range, the hypothetical cloud and coat look less flat | Keep the revised input only if the intended appearance improves; the manual does not promise this result |
| Interpolation comparison | Tetrahedral and trilinear appear similar in this hypothetical image | Leave interpolation as an independent choice; this observation says nothing about range repair |
Limits and what to try next
The manual defines the boundary: values outside [0,1] are clipped. It does not specify a universal adjustment, threshold for diagnosing clipping from appearance, or guarantee that reducing the input range will preserve a particular highlight or shadow. That is why the comparison above is a controlled author recommendation rather than a documented automated test.
Interpolation calculates colors for inputs that do not land exactly on a LUT cube node. darktable offers tetrahedral, trilinear, and pyramid methods; the manual says differences are usually not visible except with smaller LUTs. These methods do not extend the LUT’s input range, restore detail removed earlier, or recover information absent from the source. If the color space and input-range comparison do not explain the result, verify the LUT’s intended compatibility and workflow with its documentation. Do not treat a LUT’s advertised look as proof that it suits every darktable pipeline.
Common questions
Does a flat highlight prove the LUT clipped values? No. It is a reason to investigate; appearance alone does not establish that the input was outside [0,1].
Should I always lower the input range? No. Try it as a comparison when out-of-range input is plausible, then judge the intended image. It is not a universal correction.
Can another interpolation method bring back lost texture? No. It changes how between-node colors are calculated, not the input range or missing image detail.
Sources: darktable 5.6: LUT 3D describes clipping, input types, LUT use, color-space selection, and interpolation. Its note that camera-log conversion with this module has not yet been tested is relevant when identifying that distinct use case.
Sources and next step
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.