Haldi is the one ceremony where yellow is the entire point, and also the thing that destroys your grade. Here is how to hold the pigment without turning every face on screen into a turmeric smear.
Most ceremonies give you one hard color problem. Haldi gives you two, stacked on top of each other: the turmeric itself is a saturated, warm yellow that your camera clips before you can blink, and the subject wearing it has skin tones that sit uncomfortably close to that same hue on every vector scope you own. Pull the yellow down to save the skin, and you lose the ceremony. Leave the yellow alone, and every face turns into a highlighter pen.
The fix is surgical. Not a LUT, not a curve. A qualifier. But you have to build the qualifier correctly, or you will spend an hour chasing it in circles. Let's go through the exact setup.
Understanding the Problem in Scope Terms
Before you touch a node, pull up a vectorscope and look at your raw footage. On S-Log3/S-Gamut3.Cine material from an FX3 or A7S III (the cameras you are most likely grading if you shot outdoors at a morning haldi in natural light), the yellow turmeric paste will sit at roughly the 11 o'clock position on the vectorscope, pushing toward the yellow-green boundary. Skin tones on a mid-brown South Asian complexion will cluster along the classic skin tone line, which runs roughly from the center to about 10:30.
That overlap is your entire problem, stated in three sentences. The paste and the skin are adjacent in hue. A global saturation reduction hits both. Any yellow curve adjustment that is wide enough to catch the paste is wide enough to desaturate the forehead.
The correct reading on a well-graded haldi shot: the paste sits vivid and warm, the face underneath holds texture and readable color, and the overall key does not feel like a studio strobe blast of pure lemon. You want warmth, not fluorescence. That is a different thing, and the grade has to make the distinction explicitly.
The Node Architecture
Work in DaVinci Resolve with a proper color managed pipeline. If you shot S-Log3/S-Gamut3.Cine, your first node is a Color Space Transform bringing you into DaVinci Wide Gamut / DaVinci Intermediate. From there, your second node is a primary balance: set your white point off the whites of a dupatta or any clean neutral fabric in frame, cool the shadows slightly (nothing aggressive, around -3 to -5 Offset Blue in the shadows only), and get your exposure in the right ballpark.
Your creative or LUT node comes third. If you are using a pre-built grade for haldi work, apply it here as a node-based LUT via Node LUT, not a timeline LUT, so everything downstream still operates in a clean color space.
The fourth node is where the haldi-specific work happens. This is a Qualifier node, and its only job is to protect skin from what you are about to do in node five.
Building the Skin Qualifier the Right Way
In node four, open the HSL Qualifier. Sample the skin directly from the face, not the paste. You want to isolate the actual dermis: cheeks, forehead, chin. The hue range you are selecting is roughly 15 to 45 degrees on the hue wheel (warm orange-red range). Keep the saturation floor around 0.12 so you are not pulling in desaturated highlight rolloffs.
Now invert the matte. You are not going to grade the skin; you are going to use this qualifier to protect it. Feather the edges with a Blur Radius of 4–6 pixels. The result: everything outside the skin selection is available for adjustment, and the face itself is masked out.
In node five (connected in parallel to node four, not serial), pull a Hue vs Saturation curve in the yellow band. Bring saturation down on that yellow peak by about 30–40 percent. Watch the paste. It will soften from a fluorescent scream to a warm, readable smear. Then pull the Hue vs Hue curve on yellow and nudge it slightly toward orange, maybe 5–8 degrees clockwise. This small rotation separates the paste hue from the skin hue just enough to give your qualifier cleaner edges on the next pass.
The skin, masked off by node four's qualifier, should be completely unaffected. Scrub through a few frames where the subject moves into direct sun and check that the mask tracks. If it does not hold in highlights, tighten the luminance ceiling on your qualifier down to about 0.75. That drops specular skin highlights out of the selection but the forehead mid-tones stay protected.
The Luminance Problem Nobody Mentions
There is a second issue that appears specifically in midday outdoor haldi shoots, which is almost every haldi shoot. The ceremony typically happens in the morning or late morning, often in a courtyard or on a terrace: think Rajasthan havelis, Kerala homestays, Mumbai apartment terraces. That means you have harsh overhead light bouncing off a subject who is covered in reflective wet turmeric paste.
The paste itself holds luminance differently than skin. In Log footage, the paste in direct sun will push into the upper third of your Parade scope in the green and red channels simultaneously, which reads on screen as a near-white smear that barely registers as yellow at all. This is different from the paste in shade, which holds its color beautifully.
For the blown highlights: these are mostly unrecoverable, and you should not spend time trying. The grade's job in those areas is to hold the rolloff cleanly rather than introduce a green or magenta fringe as the highlight clips. A gentle Highlight Saturation pull of -15 in your primary node (node two) handles this without touching the mid-range where the paste's real color lives.
Indian Wedding LUTs
Pre-built grades for haldi, sangeet, and mandap sequences, each balanced for South Asian skin across Sony Log, Canon Log, and Blackmagic Film sources.
Camera-Specific Notes
If you are cutting together material from multiple bodies, which is common for haldi because everyone is moving, everyone is throwing powder, and you have a second shooter somewhere in the chaos, you will have a mismatch to manage before the qualifier even makes sense.
FX3 / A7S III in S-Log3: The yellow hue on these cameras in daylight tends to sit slightly green-heavy before CST. After your input CST, it resolves cleanly. The qualifier described above works as written.
Canon R5 / R5 C in Canon Log 3: Canon's yellow renders slightly warmer out of the CST, sitting closer to orange. You will need to shift your Hue vs Saturation control band about 5–10 degrees counterclockwise to catch the paste properly. The skin tone separation is actually a little easier here because Canon skin tones sit more firmly in the red-orange zone.
BMPCC 6K in Blackmagic Film: The most saturated yellow response of the three. Plan to bring the yellow saturation down more aggressively, closer to 45 percent reduction, and watch for a slight magenta cast that can appear in the paste mid-tones after the correction. A Hue vs Hue nudge back toward pure yellow fixes it in two clicks.
Drone footage (Mavic 3 in D-Log M): If you have a wide establishing shot from above during the ceremony, the paste loses its color complexity quickly at altitude because you are resolving it into a flat blob of color. Do not try to match it to the close-up grades precisely. A simple luminance cap and a -20 Saturation in Highlights keeps it readable without the absurd fluorescence that D-Log M can produce in yellows. For more on aerial color work, the Jaisalmer desert haze post covers the D-Log M tone structure in more detail.
Skin Tone Sanity Check
After the qualifier and the HSL corrections are in place, go back to your vectorscope with the skin tone line visible. Your subject's face should cluster on or close to that line. If the forehead is sitting more yellow than the cheeks, your qualifier is leaking: tighten the hue range or increase blur to smooth the matte edge. If the skin looks desaturated and the person looks slightly grey, your qualifier inversion is not working correctly, or node four and node five are wired serial when they should be parallel.
A final Hue vs Saturation boost on the red-orange range, narrow and moderate, maybe +12 to +15 percent, brings back any warmth that the yellow correction incidentally pulled out of the skin. Keep it narrow. You are not boosting skin globally; you are restoring what a wide correction took.
The result should pass a basic human test: a frame from the middle of the ceremony should look like a warm, joyful, slightly messy morning ritual, not a product shot for a turmeric supplement brand. The paste should read as tactile and pigmented. The subject's face should read as a person, not a surface.
Where LUTs Fit into This Workflow
A correctly designed wedding LUT helps here precisely because the best of them are built with this problem in mind. A LUT that has been tested on actual haldi footage will have baked-in restraint in the yellow-green zone of its color gamut mapping. You are still going to need the qualifier work described above for extreme paste coverage, but your starting point is already in the ballpark rather than two stops of correction away from it.
What a LUT cannot do is distinguish between the paste and the face. That is always a manual qualifier step, and it is always worth the ten minutes it takes to build it right. The grade that makes a haldi sequence feel real is the one where the color does exactly what it looks like it should do on the day, which is vivid and warm and specific to the ceremony, not generic and hot and digital.
For a broader look at how color space transform order affects all of this, see the CST order post, which covers why the input transform placement changes everything downstream. And if you are building out a full wedding color pipeline beyond just this ceremony, the wedding LUT pack covers every main sequence from haldi through vidaai.