How to Fix Mixed-Lighting White Balance in Portraits

A portrait made under a warm lamp and a cool window often has no single correct white balance. Neutralize the lamp and the window side may turn blue; correct the window and the room can become unpleasantly orange. The useful goal is not to make every surface neutral. It is to make the face believable while allowing the environment to retain intentional color contrast.

Mixed lighting is a local-color problem because different parts of the scene receive light with different color characteristics. A whole-image correction can improve one region while worsening another, which is why region-specific correction is often necessary. The principle is covered in this white-balance primer and in Cornell’s work on white balance under mixed lighting.

Begin with the light that defines the face

Before moving a slider, identify which source contributes most to the face. Look at the forehead, cheeks, nose and jaw rather than the background. Is one source dominant, or does a visible boundary divide warm and cool light across the skin?

Choose the dominant facial light as your starting reference. This does not mean that all skin must have exactly the same hue. A cheek turned toward a window can remain cooler than the side facing a tungsten lamp. What matters is that the variation reads as light falling across a person, not as two unrelated colors pasted onto the face.

Keep the original raw file or an untouched master before editing. In Lightroom and Camera Raw, retain the original and work through reversible settings. In Photoshop, use adjustment layers and masks rather than applying direct Image > Adjustments commands to the image layer. Adobe distinguishes these approaches because direct commands alter the layer, while adjustment layers preserve editability.

Set a restrained global baseline

Use the global white-balance controls to bring the dominant facial light into a reasonable range. In Lightroom Classic, Temperature moves the image along the yellow-blue axis, while Tint addresses the green-magenta axis. Raw and DNG files provide broader white-balance options than rendered JPEG, TIFF and PSD files, whose controls are relative and more limited.

Do not force perfect neutrality at this stage. Move Temperature until the skin stops looking obviously chilled or overheated, then use Tint to remove a distracting green or magenta bias. Watch the whites of the eyes, teeth and neutral clothing, but do not demand that they become laboratory-gray references. Reflections, makeup and surrounding colored surfaces can all influence them.

A good baseline usually leaves the second source visibly warm or cool. That is expected. Automatic white balance may average competing sources rather than reproduce how the scene felt, and some temperature difference can be an intentional photographic choice. Judge the portrait visually instead of treating an automatic result as a neutral truth.

Mask the second light, not the entire portrait

Once the dominant light is under control, create a mask for the region affected by the second source. Adobe’s current Lightroom Classic masking documentation includes People, Subject, Background, Brush, Linear Gradient, Radial Gradient, Color Range and Luminance Range tools. People masks can isolate skin or hair, and Add or Subtract controls can refine an automatic selection that has wandered into clothing or background.

Choose the mask according to the shape of the light. A Linear Gradient suits window light fading across one side of the frame. A soft Brush is better for an irregular patch across a cheek or jaw. Color Range can help constrain a correction to a particular cast, while Luminance Range can keep it away from highlights or shadows that should remain untouched.

Within the mask, adjust Temperature and Tint in the opposite direction of the unwanted cast. Make smaller moves than you think you need. If one side of the face is orange, cooling it until it becomes identical to the window-lit side can flatten the lighting and remove the scene’s atmosphere. Aim for related skin color, not cloned skin color.

Protect transitions around hair and facial edges

The quality of the mask matters most where the two sources meet. Hair, ears, jawlines and translucent fabric reveal crude corrections quickly. A hard mask boundary can create a pale rim, a green halo or a suspicious stripe across the cheek.

Use generous feathering where light blends gradually. Adobe describes Linear Gradients as broad adjustments with a progressive fade, while brush and radial-mask feathering can soften local transitions. If a People mask includes hair that should retain a cool rim light, subtract the hair or refine the boundary with a soft brush rather than warming the whole selection.

Inspect the mask at 100% view. At that magnification, one image pixel is represented by one monitor pixel in Photoshop, making small halos and hard boundaries easier to see. Mask overlays, including black-and-white views in Camera Raw, can also expose selected background, missed gaps and contamination around hair.

Use local saturation only after temperature and tint

A color cast sometimes looks severe because one channel is highly saturated, not because the temperature is entirely wrong. Correct Temperature and Tint first. If the affected area still looks unnaturally orange, blue or green, reduce Saturation slightly within the same mask. Lightroom and Camera Raw provide local Hue and Saturation controls alongside local white-balance adjustments.

Avoid draining all color from the skin. Skin contains red, yellow and sometimes cool reflected notes that give it dimension. Heavy desaturation produces a gray patch that may be technically quieter but visually lifeless. If Hue is used, keep the shift narrow and compare it with adjacent skin at normal viewing size.

Know what the correction can and cannot recover

White-balance and masked color adjustments reinterpret captured color values. They can reveal more convincing relationships within information already present, but they do not reconstruct missing texture or synthesize detail that the camera failed to record. A severely clipped or nearly featureless channel may leave little usable variation to balance.

That distinction matters when the portrait begins to invite more intervention. Reversible color and tonal work changes how existing information is rendered; painting new skin texture, replacing facial detail or generating content is reconstruction. Keep those operations separate and clearly layered if the image genuinely requires them.

Finish at normal viewing size

After the 100% inspection, zoom out to the size at which the portrait will actually be seen. Ask whether the face attracts attention naturally, whether warm and cool areas still describe the light, and whether any corrected patch looks smoother or duller than its surroundings. Pixel-level neatness is not enough if the portrait loses its atmosphere from across the room.

For another way to think about preserving facial shape while controlling distracting color relationships, the same visual discipline carries into black-and-white portrait conversion. In either case, the face needs coherent light more than it needs indiscriminate uniformity.

My stopping point would be a believable face with a trace of both sources still visible. Let the window remain cool. Let the lamp keep some warmth. White balance should organize the color for the viewer, not scrub away the evidence of where the portrait was made.

Frequently Asked Questions

Can one global white balance fix a mixed-light portrait?

Usually not. Different regions may be lit by sources with different color characteristics, so a global correction can improve the face while pushing the surroundings toward an unwanted cast. Establish a restrained baseline, then correct the second light locally.

Which mask works best for correcting the second light source?

Match the mask to the light. Use a Linear Gradient for a broad, gradual wash; a soft Brush for an irregular patch; and Color Range or Luminance Range when the correction needs tighter color or brightness limits.

Why do halos appear around hair and jawlines?

They usually come from a mask that is too hard, too broad or contaminated by the background. Refine it with feathering and Add or Subtract controls, then inspect the boundary at 100% and with the mask overlay visible.

Can white-balance correction recover missing skin detail?

No. Temperature, Tint and masked color adjustments reinterpret captured values; they do not rebuild clipped texture or synthesize detail the camera never recorded. Keep any reconstruction separate from reversible color work.