Before changing global power in darktable’s Color Balance RGB, set and check the white fulcrum in the Masks tab. In a scene-referred pipeline, white is typically greater than 1 rather than fixed at display-referred 100%, so the fulcrum normalizes the unmasked power operation. Use the picker on a deliberately chosen bright region, record a baseline, then compare a restrained adjustment and review filmic RGB highlight recovery.
Choose a sample and establish a baseline
The manual describes global power as a constant RGB exponent applied without a luminance mask. An exponent behaves differently above and below 1, so the manual says this operation needs normalization in an unbounded scene-referred pipeline. Its normalization control is white fulcrum in the Masks tab. This is distinct from contrast gray fulcrum, which sets the neutral point for the Master tab’s contrast control. See the Color Balance RGB manual.
For a window-side RAW photograph, start with these steps:
- In the Masks tab, inspect the current white fulcrum before touching global power. Note whether it is still at its default; the manual warns that leaving the default may cause artifacts later in the pipeline.
- Identify a bright region that represents the scene’s intended white point. As an author’s selection recommendation, a broad, textured, non-specular surface may be preferable to a tiny reflection or light source when anchoring an ordinary illuminated object, because it reduces the chance that a point maximum dominates. This is not a darktable requirement: the official picker uses the maximum luminance in whichever region you select.
- Use the white-fulcrum picker on the chosen region. The documented behavior is to set the fulcrum to the maximum luminance from that selected region; do not assume it calculates an average or identifies a neutral color.
- Record the chosen region and inspect a baseline: note visible highlight texture, the brightness of a middle-tone subject, and any existing highlight-recovery appearance. Then make a small global-power adjustment and compare those same areas.
The picker can only use the area you choose. If a bright region contains a specular glint or direct light source, its maximum may be a poor anchor for the white you want to normalize against. The manual explains the picker behavior and the scene-referred reason for normalization in its thresholds section.
Hypothetical acceptance worksheet
This is a hypothetical review, not a software test or a promised result. Imagine a window-side portrait with textured white curtains, a small reflected glint, a face in the mid-tones, and filmic RGB highlight recovery enabled later in the processing pipeline. The observations below are comparison prompts, not expected darktable output.
| Check | Hypothetical observation | Decision |
|---|---|---|
| Candidate sample | Curtains show texture; a separate pin-sized glint is brighter | Pick the curtains if they represent the intended scene white; do not let the glint choose the anchor by default |
| Before power | Curtain folds and facial mid-tones are both distinguishable | Record these as comparison areas; keep the fulcrum choice separate from a desired stylistic change |
| After a restrained change | Face brightness shifts and the curtain texture looks less distinct | Treat this as an observed tonal change; compare with the baseline before deciding whether the adjustment suits the image |
| Filmic RGB review | Highlight recovery now looks unexpected | Flag it for diagnosis; do not conclude from this observation alone that the fulcrum caused it |
Acceptance means you can name the sample region, explain why it represents the intended white, and describe the visible changes in both highlights and mid-tones. It does not mean the picker or a particular power setting guarantees a preferred rendering.
Limits and what to do when it fails
Global power is not a local brightness control: it has no mask in this module. If only the curtains or face need a local change, do not expect global power to isolate that region. The manual describes luminance masks for other shadow, mid-tone, and highlight controls; those masks are a separate mechanism from global power.
If highlights look wrong, compare with the recorded baseline and check whether the selected region was dominated by a point reflection or light source. Reassess the white fulcrum with a more representative region, then review filmic RGB highlight recovery separately. The manual identifies that downstream module as a prominent place where artifacts may appear when the default fulcrum is left unchanged; it does not establish that every highlight anomaly has that cause. Also consider the rest of the pipeline: this workflow cannot recover clipped detail or prove that an unknown bright surface is physically neutral.
Common questions
Should I use the brightest pixel in the whole image?
No automatic whole-image choice is implied. The documented picker uses the maximum luminance in the selected region, so a lamp or reflection may not represent the white you intend to normalize.
Is white fulcrum the same as contrast gray fulcrum?
No. White fulcrum normalizes global power and brightness levels. Contrast gray fulcrum sets the luminance left unchanged by the Master tab’s contrast adjustment.
Does a correct fulcrum guarantee filmic RGB will recover highlights?
No. The manual warns about possible downstream artifacts, but does not promise recovery or attribute every anomaly to this setting. Use the fulcrum as a normalization check and evaluate the actual rendering.
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.