Banquet hall tungsten doesn't just shift your white balance — it saturates deep into the red-orange channel and eats all the nuance out of lehenga embroidery, dupattas, and anything cream or gold. A white balance slider alone won't fix it. Here's the full correction pipeline.
You shot the sangeet at 3200K, you set your white balance in-camera to match the venue's tungsten wash, and you still pulled the card at the end of the night with footage that looks like it was captured inside a jar of honey. You're not wrong. You're not imagining it. The problem isn't your white balance — it's the physics of what tungsten emission actually does to your sensor, and why a single-axis correction can't fix a multi-axis problem.
Why Tungsten Behaves Differently from a Neutral Color Shift
Most color temperature problems are relatively well-behaved: the whole image slides cool or warm, and a corresponding shift along the blue-orange axis of your color wheel brings it home. Tungsten at banquet hall intensity is not that. A hard tungsten wash — typically 50 to 200 par cans running at 2700–3200K at full intensity over a reflective marble floor — doesn't just push the image warm. It selects for red and orange, amplifies them, and clips into yellow where red and green overlap on the sensor. Meanwhile, the blue channel starves.
The result in practice: your white correction lifts the blue channel and removes the gross amber cast, but now your reds are still over-saturated, your skin tones have a red-magenta bias on the cheeks, your cream and off-white lehengas have gone slightly green-yellow (because you've lifted blue into a scene that still has red-orange dominance), and your golden embroidery — which should be warm but not blown — is sitting at a hue angle that reads muddy. You've solved the temperature problem and created three new ones.
The fix is not a single correction. It's a four-node pipeline: a global temperature-tint correction, a targeted red-channel desaturation, a skin-tone qualifier holding your faces, and a final luminance curve that opens up the shadow detail that tungsten always crushes.
The Node Pipeline: Built for S-Log3 and S-Cinetone Alike
This pipeline is described for DaVinci Resolve's node editor, but the logic maps identically to Premiere's Lumetri or FCPX's Color Board if you understand what each step is doing. Footage examples below assume S-Log3/S-Gamut3.Cine from an FX3, FX30, or A7S III — the most common combination you'll encounter at Indian wedding receptions. The same logic applies to C-Log3 from an R5 C or C70; the channel weights are slightly different but the structure is identical.
Node 1 — Color Space Transform (CST)
Your first node is a Color Space Transform OFX from S-Log3/S-Gamut3.Cine to DaVinci Wide Gamut / DaVinci Intermediate. Do not convert directly to Rec.709 yet. Working in DWG at this stage gives you enough gamut room to move the reds without clipping. If you convert to Rec.709 first, you'll be operating on already-compressed data and your corrections will posterize. Set tone mapping to None — you're managing exposure yourself.
Node 2 — Global Temperature + Tint Correction
On your second node, open the Color Wheels panel and dial in your primary correction using the Lift / Gamma / Gain wheels. For a hard 2700K tungsten wash with no ambient fill, you'll typically need somewhere around Gain shifted toward blue by +0.04 to +0.06 on the blue channel (use the channel bars below the wheel, not the global wheel drag — precision matters here). Tint correction: if the venue runs warm fluorescent accents alongside the tungsten, you'll see a green push in the shadows; counter with a mild magenta offset in Lift of +0.01 to +0.02. If it's pure tungsten, leave tint alone at this stage.
Check your correction against a white or near-white surface in frame — tablecloth, someone's white kurta, the stage scrim. Your scopes should show the RGB parade balanced within about 15–20 IRE of each other on that surface. Don't chase perfect — you're setting a foundation, not finishing.
The Red Channel Is Still Lying to You
This is where most colorists stop and wonder why the result still looks "wedding-video orange." Your global correction has centered the image, but the red channel absorbed more photon energy during the shoot. It's still running hot relative to green and blue. You need to address it directly.
Node 3 — Targeted Red Desaturation via Hue vs. Saturation
Add a new serial node. Go to Curves → Hue vs. Saturation. Find the red band (roughly 0–30° and 330–360° on the hue wheel). Pull the curve down by about 15–25% in the red zone only. Watch your scopes — specifically the vectorscope — as you do this. You want the skin tone indicator line to drift toward neutral, not cross it. If you go too far, skin goes gray-green. Typical target: land the skin tone cluster sitting on or slightly warm of the skin-tone line, not straddling it hard toward red.
Now do a secondary check on the orange-yellow band (30–60°). Tungsten almost always pushes lehenga gold tones into a muddier orange. A very slight saturation lift of 5–8% in the yellow band (50–70° on Hue vs. Saturation) recovers the "metallic" quality of zari embroidery that the tungsten flattened. Don't confuse this with a luma lift — it's saturation recovery, not brightness.
Node 4 — Skin Qualifier + Face Hold
Add a new parallel node (not serial — you don't want this affecting the rest of the image). Use the HSL Qualifier to isolate skin tones. In tungsten-corrected footage, South Asian skin usually lands between 15° and 30° hue, saturation 30–55%, and luminance 30–65%. Pick a mid-cheek sample, then use the Matte Finesse controls to tighten the selection: raise the Clean Black slider until the tablecloth and warm-colored clothing fall out of the matte, and nudge Blur Radius to about 0.8–1.2 so edges don't strobe. Apply a gentle Hue Rotation of –2 to –4° toward yellow in the face node — this counters the slight magenta push that node 2's blue-lift almost always introduces into warm skin.
Add a Highlight rolloff of +3–5 on the skin qualifier node's saturation to gently hold blown-out cheeks from going nuclear on close-ups with direct stage lighting.
Indian Wedding LUTs
Pre-built nodes for exactly the tungsten-LED-fire mixed lighting you face every sangeet — designed for Sony S-Log3 and Canon C-Log3 footage shot in Indian banquet halls and wedding venues.
Shadow Recovery and Final Output Conversion
Tungsten at full intensity is not just a color problem — it's a contrast problem. The sheer volume of emission light collapses your shadows. The dance floor at a sangeet routinely runs 6–8 stops of dynamic range from lit face to unlit floor, and your tungsten correction, by lifting the blue channel, may have opened up some noise in those shadows. You need to address this before your output CST.
Add a node after your corrections. In Curves → Custom, place a control point at roughly 20 IRE and lift it by 8–12 IRE. Then place another at 35 IRE and lift by 3–5 IRE. This creates a gentle shadow toe that pulls up the crushed dance floor detail, opens up the dark backgrounds behind the couple, and makes the whole scene feel lit rather than spotlit. Watch your blacks — don't let them go milky. A scopes check: your Parade shadow floor should stay at or below 5 IRE.
Final node: a second Color Space Transform from DaVinci Wide Gamut / DaVinci Intermediate to Rec.709 / 2.4 gamma (or your delivery spec). Tone mapping: Luminance Mapping at default, unless you've managed a hard overexposed highlight area manually in a prior node — in that case, set to None and control the roll-off yourself in the curve node above.
When You Have Mixed Sources: Tungsten Plus DJ LEDs
A note for the shoots where the DJ lights are running simultaneously — which is most of them after 10 PM. The multicolored LED wash introduces a problem your static correction can't fully address: different frames, different casts. Here your node correction becomes a starting point, not a finish.
The approach: set your tungsten correction as a base on the timeline clip, then use Remote Grades in Resolve to create per-shot variations. For shots dominated by the tungsten wash, your base grade applies clean. For shots lit primarily by a green or blue DJ LED throw, add an adjustment clip or a node version switch that dials back your blue-channel lift (because you no longer need it) and pushes saturation back up slightly in the cool range. The principle is that your base grade is built for the dominant light source, and your per-shot adjustments handle the outliers — you're not building six separate pipelines, you're building one and patching exceptions.
This is the same workflow described in detail for a related multi-source problem in the post on grading mandap fire without blowing the lehenga — worth reading in parallel if you're doing full-ceremony grading in a single session.
The Camera Decisions That Make the Grade Easier
Before you ever open Resolve: your in-camera choices at the shoot set a ceiling on how well this correction works. Three rules that matter specifically for tungsten-heavy venues:
- Shoot S-Log3, not S-Cinetone, for coverage in fixed tungsten light. S-Cinetone is an excellent profile but it has already baked in a partial color render. At 2700K, that bake can compound the tungsten cast into the orange in ways that are harder to unwind than clean log. Log gives you the full data range to work with.
- Manual white balance, not Auto. Auto WB on the FX3, A7S III, and R5 C will breathe through a sangeet whenever the DJ lights pulse. Your grade is going to be built on a static correction — you need consistent raw data coming in, not per-frame WB shifts.
- Set your in-camera WB to 3200K. Not 2700K (even if the venue runs 2700K). The extra 500K headroom toward daylight reduces how hard you need to push the blue-channel lift in post, which in turn reduces noise amplification in the shadows. You're not trying to nail it in-camera; you're trying to land the raw data in a range your grade can work with cleanly.
For a deeper look at how different Sony bodies handle skin tone bias before correction — including why the FX30 trends cooler than the FX3 under identical conditions — the post on FX30 vs. FX3 skin tone behavior is directly relevant to this workflow. The per-body baseline affects how much you'll need to adjust the HSL qualifier ranges above.
The sangeet is consistently the hardest technical environment in the Indian wedding day — harder than the mandap fire, harder than the haldi's flat overcast. It's hard not because the equipment fails but because the light source is designed for spectacle, not for cameras. Understanding exactly what the tungsten emission does to your sensor data is the prerequisite to fixing it. The pipeline above is your map through it.