Reception Under Mixed Daylight, Tungsten and LED
A wedding reception is being photographed in a covered courtyard in Lucknow in November. The shamiana sides are open to late-afternoon daylight, while the couple stands under a canopy lit by 40 W tungsten bulbs, a warm-white LED string and the videographer’s LED panel. The camera is a full-frame body with a 24–70 mm f/2.8 lens at ISO 1600, white balance set to 5500 K, and JPEG capture. On the LCD the bride’s cream lehenga looks acceptable, but the groom’s white kurta has a yellow-green cast and his skin looks sallow. An incident meter at the groom’s face reads f/2.8 at 1/60 s, ISO 1600; a spot reading on the white kurta is 1.5 stops higher. The on-camera speedlight is set to TTL, and the first frames show a hard shadow behind the couple. The client wants 60 edited images by the next evening.
1. Explain why the white kurta and the skin show different colour casts, and state what should be changed at capture to make the next event more controllable.
Read the worked answer
The scene contains at least three light sources with different spectral power distributions: daylight near 5500 K, tungsten bulbs near 2800–3000 K, and a warm-white LED string whose spectrum may be strong in yellow-green and weak in red. A single camera white balance cannot neutralise all of them at once. At 5500 K the tungsten-lit areas are rendered orange-yellow, while the LED may add a green component. The white kurta reflects the mixture strongly across the spectrum, so it shows the combined cast; skin has lower reflectance in the blue-green region and higher red, so the same light makes it look sallow rather than green.
The deeper problem is JPEG capture: the white balance is baked in, leaving little correction room. At the next event I would shoot raw, place a grey card at the subject position under the key light, and set a custom white balance from it. I would decide which source should be neutral - usually the light on the faces - and gel the flash to match it: CTO on the flash if keeping tungsten ambient, or CTB on daylight openings if bringing everything to daylight. The videographer’s LED panel should be gelled or switched off during stills. I would also check the RGB histogram, not the LCD, and keep a colour checker in the first frame.
2. Using the meter readings, propose an exposure and flash-balance plan that holds the white kurta and gives a softer shadow. Show the stops.
Read the worked answer
At ISO 1600 the face reads f/2.8 at 1/60 s. The white kurta is 1.5 stops brighter. If I shoot at f/2.8, the kurta sits only about half a stop below the JPEG clipping point, which is risky with a white fabric that has texture. Stopping down to f/4 puts the face one stop under the metered value and the kurta at +0.5 stop relative to the face, which is much safer and still within the sensor’s highlight headroom. I would keep 1/60 s, which is within normal flash sync and slow enough to collect ambient.
The ambient face is now one stop underexposed. The flash should fill the face back towards f/4, not blast it. With TTL I would start at about −0.7 to −1.0 EV flash compensation and test: the face should sit slightly above ambient, while the kurta remains below clipping. The hard shadow comes from the small on-camera source. Bouncing off a white ceiling or wall, or moving the speedlight onto a bracket with a small umbrella or softbox, enlarges the source and softens the edge. If the ceiling is coloured, I would use a bounce card rather than accept a colour cast. I would also check that no background guest is within a metre of the couple, because the shadow is most visible there.
3. Outline the post-production correction for the already-shot JPEGs and explain which problems cannot be fixed.
Read the worked answer
The JPEGs are 8-bit and already white-balanced, so correction must be conservative. I would open each file and set a neutral reference on the grey card or on a neutral part of the groom’s kurta, then use a curve to reduce the green in the midtones where the LED cast is strongest. Skin is better handled with a local mask and a small hue/saturation adjustment than with a global magenta shift, which would turn the cream lehenga pink. Background areas lit mainly by tungsten may need a separate warmer or cooler correction from the faces; one global white balance will not fix both.
What cannot be recovered: blown highlights on the kurta if any channel reached 255, and the separation between flash-lit foreground and ambient background once they are merged in JPEG. Heavy correction will also expose 8-bit banding in smooth shadow areas. I would correct the strongest casts, keep skin natural rather than plastic, export sRGB JPEGs for the client, and explain that the next event will be shot raw with a grey-card reference. I would not promise that every frame can be made neutral.