The agni kund is burning, the bride's silk lehenga is catching every lick of orange, and your waveform is a vertical wall on the right side of the frame. Here is how you hold both — without desaturating your way into a grey ceremony.
You're sitting in Resolve looking at a shot from the mandap — priest gesturing over the fire, bride and groom seated close, the lehenga a deep crimson or gold-embroidered maroon. It's one of the most important frames of the day. It's also, optically, a disaster: an open flame in the near-mid ground, two subjects lit almost entirely by that flame and whatever ambient light the venue allowed, and a camera that had to commit to a single exposure for all of it.
The waveform tells the story. Flame core: clipped. Silk highlights: 95–100 IRE, teetering. Faces: mid-50s if you were conservative, low-40s if you chased exposure on the fire. Every move you make to recover the flame blows the lehenga. Every move you make to protect the lehenga flattens the fire into muddy orange.
This is a tonal separation problem, not a LUT problem. No single LUT resolves it because the problem is structural — the dynamic range you need doesn't exist in the image unless you extract it carefully before the LUT ever sees the footage.
Why the Lehenga and the Flame Are Locked Together (and How to Unlock Them)
Crimson silk and open flame share an address in the color gamut: both are heavily weighted in the red channel, both are warm, both have high saturation. When you pull the highlights down to recover the flame, you're pulling the same region of the tonal range that holds the lehenga's deep reds. The fabric flattens, the embroidery disappears into the base cloth, and the whole frame loses the ceremonial weight it needs to carry.
The unlock is tonal separation via luminance qualification. You're going to split the image into at least two separate tonal regions — the super-highlights (flame) and the high-upper-mids (fabric) — and treat them independently. This is only possible if your source codec has the range to support it. Anything above S-Log3/S-Gamut3.Cine on the FX3/FX6/A7S III, Canon Log 3 on the C70/R5 C, or BMPCC 6K's raw track gives you enough. If you shot in-camera picture profiles at 8-bit 4:2:0, you have less headroom — dial back your ambition on the flame recovery accordingly.
The Node Tree, Node by Node
Build this left to right in Resolve. Six nodes. Each has one job.
Node 1 — Color Space Transform (Input)
If you already have a Color Space Transform or a LUT doing your input transform on a parent node or timeline-level, skip this node or consolidate. You need your footage in a working space before anything else. Input: S-Log3 / S-Gamut3.Cine (or your actual log profile). Output: DaVinci Wide Gamut / DaVinci Intermediate. Tone mapping: off — you're not tone-mapping here, you're doing that manually across the chain.
Node 2 — Pre-LUT Exposure Correction
Set your overall exposure here using Lift / Gamma / Gain or the Log wheels. Aim for faces in the 45–55 IRE range. The fire will be clipped or near-clipped — don't chase it here. This node is about getting your subjects to a workable starting point. Typical move: Gain down 0.3–0.5 stops, Gamma up 0.10–0.15 to compensate midtone crush.
Node 3 — Flame Highlight Recovery (Luminance Qualifier)
This is the key node. Add a serial node. Enable the Qualifier (HSL, but switch to Luminance qualification mode). Sample the flame core — you want to select luminance values from roughly 85 IRE upward. Set your Low: 0.75, High: 1.0, Softness Low: 0.20, Softness High: 0.0. The soft rolloff on the low end is critical — it lets the transition into the fabric happen gradually rather than with a hard edge.
In the Color Wheels for this node, pull the Gain (highlights) wheel down — start at -0.35 and watch the waveform. You're looking for the flame peak to come down to 90–93 IRE while the top of the lehenga stays where it was. If the fabric is also pulling down with the flame, your softness is too low — widen the Low Softness to 0.25 or 0.30.
Check your matte in the Qualifier viewer (Viewer → Highlight mode). The flame should be white or near-white. The lehenga should be grey (partially selected). The faces should be dark grey to black (mostly excluded). That gradient matters — it's what keeps the transition from looking like a mask edge.
Node 4 — Your Base LUT or Output Transform
Apply your creative LUT here. If you're working with a dedicated Indian wedding LUT built for warm mixed light, drop it in on this node. If you're using a CST output to Rec.709 and then a creative LUT on top, do both here or split them across two serial nodes. The point is: your creative look happens after the structural work in Node 3, not before.
Indian Wedding LUTs
Built specifically for warm mandap light, deep silk tones, and mixed tungsten-fire sources — pre-mapped so your lehenga reds don't clip while the ceremony stays alive.
Node 5 — Lehenga Saturation Recovery
After the LUT, the fabric reds may be muted — LUTs compress saturation in heavy red-orange regions to avoid illegal values. This node uses an HSL Qualifier targeting the deep reds of the fabric. Sample the lehenga directly. Narrow your Hue range: center around 15–30° (red-orange), width around 35–40°. Set Saturation range: Low 0.25, High 1.0, to exclude the low-sat skin tones nearby. Luminance range: exclude the very top (clipped flame) and the very bottom (shadows).
Typical moves on this node: Saturation up +15 to +25, Hue rotate slightly negative (toward red, away from orange) if the LUT has shifted the fabric toward the flame color — usually -4 to -8°. Check with parade scope: your red channel should show detail and texture, not a flat shelf.
Node 6 — Output Limiter
Add a final serial node. Open Color → Curves → Custom. Pull the top-right anchor down to 0.95 (or use Soft Clip under Video Scopes → Soft Clipping). This prevents any over-bright edge pixels from slipping through to delivery and is especially important if you're doing a separate color-managed export for a client preview versus a full-grade master. It also protects the YouTube or Instagram encoder from over-compressing the flame area.
Reading the Parade Scope to Know When You're Done
The parade scope tells you more than the picture ever will on a grading monitor. For a well-graded mandap fire shot, you want to see:
- Red channel: peak around 90–96 IRE, showing internal detail (individual embroidery threads visible as texture, not solid fill), with the flame sitting at or just below 100.
- Green channel: flame should show through here too — if the green channel is hard-clipped while red isn't, your flame will look unnaturally monochromatic. Let it breathe to 90+.
- Blue channel: low in the mids and shadows (this is a warm scene), but not zero. If blue is completely crushed, skin tones will look artificially orange. Keep some blue in the face region — typically 30–45 IRE on the skin area.
If your red and green are both sitting at 100 IRE with no internal detail visible, you need more pull in Node 3. If the red is recovered but the green is also pulling down (the flame looks greenish), your luminance qualifier is selecting too wide — tighten the High Softness.
Specific Situations That Change This Workflow
The Lehenga Is White or Ivory
White and cream lehengas are actually harder than crimson in this context because the fabric is in the same luminance neighborhood as the flame. The qualification approach still works, but you need a second qualifier layer in Node 5 targeting fabric luminance (typically 65–80 IRE) rather than saturation — use the Window qualifier to isolate the dress spatially if needed. Be cautious with Offset rather than Gain on white — it lifts all channels equally and preserves the neutral hue better than warm-biased Gain moves.
The Camera Was an FX30 or R5 at High ISO
At ISO 12800+, the noise floor in the highlights starts to interact with your qualifier. The soft grain in the fire region can cause your luminance qualifier to flicker slightly between frames. The fix: apply a Noise Reduction node before Node 3 (not after the LUT — temporal NR on a graded image introduces its own problems). DaVinci's temporal NR at Motion: 3, Luma Threshold: 8, Chroma Threshold: 12 is usually enough to stabilize the qualifier without destroying flame texture.
The Flame Has Multiple Zones
A large agni kund — common in outdoor Rajasthani weddings in venues like Jaisalmer's Sam Sand Dunes or Jodhpur palace gardens — can have a deep orange-red outer zone, a near-white core, and significant foreground smoke. Smoke picks up the warm color cast and occupies the mids in a way that competes with fabric. In this case, add a Node 3b targeting the smoke region via a drawn Power Window and use a gentle Hue vs Saturation curve to reduce the orange saturation of the smoke without touching the surrounding air or the fabric.
The mandap fire is not an enemy of the grade — it's the reason the frame is beautiful. The goal is never to eliminate it or neutralize it. The goal is to give it exactly enough room to breathe without consuming the bride alongside it. Six nodes, two qualifiers, a parade scope check, and the ceremony looks the way it felt when you were standing there. That's the job.
If you want to extend this workflow into the mixed-light reception environment — the same fire alongside tungsten chandeliers and LED stage wash — the node tree logic covered in the reception mixed-light post picks up where this one ends. And if you're newer to why log footage needs a proper input transform before any of these nodes are meaningful, the CST node breakdown covers the foundation.