Choose color proximity when the target should retain its existing color relationships with only a restrained influence from the source. Choose stronger palette dominance when the source’s relative color-cluster weights are part of the look you want to transfer. These are two ends of the color dominance control, not two separate mapping modules. Treat histogram equalization as an independent decision about contrast distribution, and change it only when the target should also resemble the source’s histogram.
Specific Judgment and Steps
The official color mapping documentation defines the module as a statistical analysis of source and target color characteristics. It does not define a universal “correct” dominance setting, so use the image pair and the intended visual relationship to choose.
- Open the proposed source image in Darkroom. Set
number of clustersto a value that roughly reflects the visible number of dominant colors, then pressacquire as source. Read the displayed source swatches and their weights as a description of the source’s sampled color groups, not as a promise that every object has its own cluster. - Open the target image after acquiring the source clusters, then press
acquire as target. The manual says the source clusters should remain stored when you open the target, but its reset instruction does not explain what reset changes. If the source clusters are missing, return to the source image, acquire it again, then acquire the target. Changing the cluster count resets all collected clusters, so acquire both images again after changing it. - Start near the color-proximity end of
color dominancewhen preserving the target’s local color character matters. Move toward the relative-weight end when the source’s dominant-versus-secondary color balance should be expressed in the target. Intermediate values are a gradual tradeoff, not a categorical third mode. - Keep
histogram equalizationlow or unchanged while judging palette transfer. Raise it only when matching the source’s contrast distribution is also an explicit goal. Adjust mapping strength separately with the module’s effect controls or normal blend mode and opacity; later modules can fine-tune color because color mapping occurs early in the pixelpipe.
A New Acceptance Worksheet
The following is a hypothetical worksheet, not a measurement of darktable’s internal algorithm or a guarantee of the result. Object names are a person’s descriptions of image content, not semantic labels generated by the module. Record actual observations after applying the mapping; the comparison prompts below are author-suggested checks, not reported test results.
| Check | Hypothetical observation | Decision |
|---|---|---|
| Check | Hypothetical prompt to fill in after applying the mapping | Acceptance check |
| Source capture | Describe visible colors in your own words (for example, blue-gray areas or warm-colored areas); these descriptions are not module-generated semantic clusters | Record the displayed source swatches and weights; do not assume a cluster corresponds to a named object |
| Target capture | Describe visible colors and note which occupy more of the image; these are human observations, not cluster labels | Record the displayed target swatches and weights |
| Dominance comparison | Compare proximity-oriented and relative-weight-oriented settings on this pair; do not predict which object will change | Check whether local target colors remain recognizable and whether the whole-image color balance moves toward the intended source relationship |
| Histogram comparison | Compare the target with equalization unchanged and, if contrast transfer is wanted, with it adjusted | Check shadows, highlights, and midtones separately for the intended overall contrast |
| Final review | Note the actual local and whole-image changes after each adjustment | Accept only if dominant-color relationships, local target colors, and overall contrast all meet the brief |
This worksheet separates three questions: whether colors remain recognizable, whether cluster weight relationships feel intentional, and whether contrast distribution has changed acceptably. The module’s swatch colors are clues; judge the actual image and its displayed type information rather than inferring behavior from a node color.
Limits and Next Steps After Failure
The documentation says cluster acquisition uses statistical sampling and may produce different results, especially with more clusters or complex palettes. Therefore, a borderline decision should be reviewed after a fresh acquisition, not treated as a precise object-to-object correspondence. The module also does not promise semantic awareness: a cluster is a statistical color group, not “the sky” or “the subject.”
If local colors look wrong but the overall palette feels useful, lower color dominance and reduce the module strength or opacity. If the palette feels right but contrast is wrong, leave dominance alone and adjust histogram equalization separately. If both are wrong, inspect whether the source and target contain comparable visual ranges, then reacquire clusters with a more suitable count. Do not claim that mapping restores detail that was absent from the target. The module accepts linear or non-linear Lab display-referred input and processes in non-linear Lab, as described in the module reference; this does not remove the need to review the image in its actual workflow.
Three Short Questions
Should dominance always be high for a strong style?
No. High dominance emphasizes relative cluster weights and can make the effect strong, but a restrained transfer may better preserve the target’s identity.
Is histogram equalization another kind of palette dominance?
No. Dominance controls how source and target color clusters are paired. Histogram equalization controls how much target contrast distribution is matched to the source.
What should I inspect first after mapping?
Inspect a recognizable local region, then compare the source-like color balance across the whole target, and finally check whether highlights, shadows, and midtones have the intended contrast relationship.
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.