Photography Lighting & Exposure

Understanding Kelvin Color Temperature and Mixed-Light White Balance

A daylight window, a tungsten lamp, and an LED practical can put three different casts on the same subject. Use a gray card at the subject plane to measure each source, then match or contain the spill before setting the final white balance. The worked setup keeps the lamp visibly warm while bringing both subject checkpoints within three RGB code values.

Overhead studio plan Distances measured to the subject Window key 1.5 m · 400 lx · 5600 K 45° A / B card marks 90° LED fill 1.2 m · 100 lx Key to fill = 4:1 Tungsten 3200 K Flag Camera axis
Read the plan from the camera axis. The window and LED contribute 400 and 100 lux respectively at the center of the subject plane, while a flag keeps the tungsten practical on the background and gray-card marks A and B reveal any remaining spill across the subject.

Choose which light sets neutral

Start with the surface that has to read accurately. For a pale ceramic piece beside a window, use the window as the key and balance to a card at the ceramic. A tungsten desk lamp can stay orange in the background. The problem starts when that lamp also paints an orange patch across the ceramic or a green LED fills its shadows. White balance applies one correction across the frame, so a setting that cleans up one patch can push another patch farther off neutral.

On a fixture, a lower Kelvin value means warmer light. On a camera or a RAW temperature control, raising the white balance setting makes a fixed exposure of the same scene render warmer. The camera is compensating for the bluer light you told it to expect. Tint compensates along the green–magenta axis. A gray card that looks green after the temperature adjustment calls for magenta correction, either at the offending fixture or in the image if the whole subject shares the cast. Record slider values with the application or camera model because units and signs vary. ARRI color FAQ

Measure the window, lamp, and LED separately

Put a matte white-balance gray card at the subject position and hold it parallel to the important surface. Mark the stand, height, and angle so the target can return to the same plane. If the card shows a sheen, tilt it just enough to lose the reflection and retain that angle for the whole series. Avoid fingers, frame edges, or the card holder casting shadows on the sample area. Lock camera position, exposure, picture profile, and white balance for the comparison. Use RAW if available.

Start with the window alone, switch to the lamp alone, then the LED alone. Use an opaque blind or a blackout panel to isolate the electric sources from the window. With a lux meter centered where the card sat, keep the sensor facing the same way as the card for every reading. Log lux at the working position and fixture output. If a source-only camera frame is too dark to sample cleanly, shoot a labeled brighter reference and keep that frame out of the exposure comparison. Use a color meter for physical CCT and green–magenta readings if you have one. Otherwise, log the RAW eyedropper's temperature and tint as camera-specific correction settings.

Example source-only readings at the center of the subject plane
SourceIncident lightRAW card correctionAction
Window key400 lx5600 K, +3 tintSet the reference
Tungsten practical160 lx before flagging3200 K, +2 tintKeep warm on the background
LED practical100 lx5300 K, +14 tintMatch the window and trim green

These are worked example values for a single RAW converter with positive tint toward magenta. Sample a clean interior patch with an averaged eyedropper, such as 11 × 11 pixels when the software offers it. Select several patches if you suspect texture, uneven illumination, or noise. A bright card with a clipped channel can produce a misleading neutral reading. Adobe Camera Raw white-balance controls

Control tungsten spill before changing Kelvin

In the example, the unflagged lamp adds 160 lux against a 400 lux window at the subject. That contribution is 40% of the key reading, enough to show on pale material. Place a freestanding flag in the path from the lamp to the subject and leave the background beam open. Use a noncombustible flag with the fixture manufacturer's heat clearance. Recheck the lamp alone at the subject. If the meter reads 20 lux after flagging, the lamp contribution is down to 5% of the key at that spot. Move the meter to the shadow side as well. A spill contribution that is small at center can dominate a shaded edge.

If the lamp must light the same neutral surface at useful strength, estimate the conversion in mireds, or reciprocal megakelvin. Divide 1,000,000 by each Kelvin reading and subtract the source value from the target. Going from 3200 K to 5600 K gives 179 − 313 ≈ −134 mired, a blue shift. Choose a heat-rated color-temperature blue gel from its published shift data, then measure again because the gel reduces output and the result depends on the source. Restoring the lost light can alter the balance between key and fill. Log the post-gel lux before judging the new mix.

Match the LED on both color axes

The isolated LED needs 5300 K and +14 tint to render gray, while the window needs 5600 K and +3. With white balance fixed at the window setting, put the card under the LED alone. Raise the LED's fixture CCT until the warm cast recedes. Then dial minus-green on the fixture or try a light minus-green gel and reshoot. Adjust one control at a time and recheck both axes after each change. Use the 11-point gap to choose the correction direction, then establish gel density or dial value by retesting.

Return the LED to the intended dimmer setting and check again. In this example the stand position, beam angle, and output are set to deliver 100 lux on the card plane against the 400 lux key. That 4:1 key-to-fill contribution ratio is a two-stop difference at the meter position. Measure the two visible sides of a curved subject separately to determine their contrast. If the neutral card matches while red fabric or skin still looks different between the sources, compare those materials in source-only frames. The sources can share a neutral point and still render colors differently. Reduce the LED contribution or swap the lamp if the mismatch matters to the shot.

Verify the mixed frame from edge to edge

Switch all three sources back on. Make one frame with the gray card at mark A on the window side and a second at mark B on the LED side, keeping the card parallel to the same subject plane. Use the window setting as the starting point and choose the final global balance from A. In the same 8-bit sRGB output preview with the same profile and tone adjustments, sample both frames and compute the spread between the largest and smallest RGB channel values. Midtone averages of 119/120/121 at A and 118/120/121 at B give spreads of two and three codes respectively. Treat three as a working tolerance for this setup and output transform.

If B reads 118/126/120, the eight-code spread and higher green channel point you back to the LED-side lighting. Turn off the LED and repeat the B frame to distinguish LED spill from a colored wall bounce. If the residual cast follows the fixture, correct it at the fixture. If it follows a reflection, block the reflection or move the subject. Retake both card frames after the fix.

  • Confirm that the card is matte, evenly lit, sharp, and free of clipped channels in both reference frames.
  • Keep A and B within the agreed RGB spread under the same output transform, using one white balance for both samples.
  • Check a real subject frame for clean edge color and preserve warmth where the tungsten beam belongs.
  • Record the Kelvin and tint, lux, dimmer level, flag position, and time. Reshoot the card when cloud cover, window angle, or fixture output changes.