Shoot a talking-head interview with the FX30 and FX3 side by side, cut them together, and skin tones will visibly drift — even when both cameras are on the same S-Log3 profile and the same white balance. This isn't a myth, and it isn't a calibration error. It's physics, sensor design, and a predictable color science gap you can correct in one node.
Pull up any multicam timeline where an FX3 and an FX30 are sharing the load — a corporate event, a documentary interview, a candid wedding second-camera setup — and you will see it the moment you slap on a standardized LUT. The FX30 renders skin about half a stop cooler, with a subtle cyan push in the upper-midrange and a slightly compressed yellow channel. The FX3 reads warmer, with more amber-gold in the flesh tones and a punchier highlight rolloff. Both cameras are shooting S-Log3/S-Gamut3.Cine. Both are set to 5600 K. Both are exposed at the same nominal value on a gray card. And yet.
Why the Gap Exists
The FX3 uses Sony's full-frame 12.1 MP back-illuminated Exmor R sensor — the same sensor architecture that lives in the A7S III. The FX30 uses a Super35 APS-C sensor derived from a different lineage, closer to the A6700. Different sensor pitch, different microlens geometry, different photosite architecture. Even when Sony maps both to the same S-Gamut3.Cine color gamut primaries, the actual spectral response curves of those two sensors are not identical. The color science pipeline compensates, but it cannot compensate perfectly, because the underlying spectral sensitivities diverge particularly in the 580–620 nm range — which is exactly where skin tone information lives.
The practical result: the FX30's sensor is slightly less sensitive to the red-orange portion of the visible spectrum relative to the FX3. Sony's in-camera matrix tries to normalize this, but the normalization is optimized for technical accuracy across the full gamut, not for perceptual skin-tone parity under the warm, mixed-source light of a real shoot. So in grading, you inherit the gap.
There is a second contributor: the APS-C crop means the FX30 sees a slightly different optical cross-section from most lenses compared to the FX3 at the same focal length and aperture. This changes the lens flare characteristics and the subtle chromatic fringing that large-format sensors tolerate differently — both of which shift the perceived warmth of the final image even at a pixel level.
Quantifying the Gap Before You Fix It
Before touching any node, measure. Open both clips in DaVinci Resolve. Go to the Color page, Color > Scopes, set the scope to Parade (RGB), and play a frame with a clean skin tone in frame — ideally the same subject, same light, same clothing. Read the midtone balance on both cameras. On a healthy FX3 grade, you will typically see the red channel sitting 5–8 IRE above the blue channel in the mid-skin register (around 50–60 IRE on Parade). On the FX30, that gap narrows to 1–3 IRE, sometimes with the blue channel even kissing the red channel. That is the cyan push made visible.
Now switch the scope to Vectorscope (YUV) and place a node that does nothing — a neutral serial node — on each clip so you can use the qualifier without affecting the output. Hover over the skin region. On the FX3, skin dots cluster just inside the skin tone indicator line at roughly 2 o'clock. On the FX30, those same dots sit slightly counterclockwise — toward the cyan-green arc. Half a step. Consistent. Measurable.
That vectorscope shift is your diagnostic. You are not imagining it. And it tells you exactly what the correction needs to do: rotate the skin cluster back clockwise — warmer, more red, less cyan — without touching the rest of the gamut.
The Single Node That Fixes It
One node, serial, placed after your Color Space Transform (so you are working in display-referred space, not log) and before any creative grade. Here is the exact build:
- Node type: Serial (standard corrector node)
- Tab: Open the Curves panel. Switch from Luma to Hue vs. Saturation — but you will not change saturation yet. First go to Hue vs. Hue.
- Hue vs. Hue — Red-Orange band: Add a control point at approximately 25–30° (orange) with a tight bell. Drag it slightly clockwise — toward red. A shift of +4 to +6° is usually the full correction. Pull back to +3° if your subject has deeper skin. Watch the Vectorscope; stop when the skin cluster sits on the indicator line.
- Hue vs. Saturation — same band: After the Hue rotation, skin can desaturate very slightly because you've moved it off the sensor's native position. Boost saturation on that same orange band by +0.05 to +0.10 (Resolve's 0–2 scale). Don't go further.
- Qualifier lock: On the same node, open the HSL Qualifier. Sample a midtone skin region. Expand the hue selection to cover the full skin range (typically 10°–55° hue range), keep luminance range between 0.25 and 0.75 to exclude deep shadows and hot highlights. Set Soft to 0.20. This confines the correction to skin without spilling into warm backgrounds, golden-hour skies, or wooden mandap decor.
- Matte finesse: Enable Blur Radius on the qualifier at 0.5. Prevents edge fringing on hairlines.
That is the entire correction. No secondary power windows. No group-level offsets. One node, six parameter touches.
If you want to verify it is clean: temporarily enable the Highlight matte overlay (the dashed square icon in the qualifier panel). Your skin pixels should glow white; background walls, fabric, and foliage should read black or deep gray. If warm tungsten highlights in the background are bleeding into the matte, tighten your luminance ceiling — bring it down from 0.75 to 0.65.
Applying It as a Node Group So Both Cameras Live in the Same Timeline
If you are cutting FX30 and FX3 footage in the same project — which is the whole point — build this correction as a Remote Grade applied only to FX30 clips, not globally. The cleanest way:
- In the Media Pool, select all FX30 clips. Right-click → Clip Attributes → confirm codec (
XAVC S-IorXAVC S). You can also tag them with a marker color in the timeline — orange, say. - In the Color page, select one FX30 clip, build the node above, then right-click the node → Copy Grade. Select all other FX30 clips (Ctrl+click or filter by marker), right-click → Apply Grade.
- Alternatively, if your project uses Shared Nodes: create the corrector node, right-click → Add Node as Shared Node. Name it
FX30_SkinMatch. Now every FX30 clip in the project gets the same correction, and if you adjust it once, all clips update. This is the right workflow for a 500-clip wedding multicam project.
The FX3 clips get no camera-matching node — they are your reference. Every creative grade builds on top of this foundation.
The Positiva Bundle
Built for multicam workflows — includes pre-matched grades for Sony S-Log3 that account for the FX30/FX3 sensor gap, so your starting point is already camera-neutral.
Where This Shows Up Most Brutally on an Actual Shoot
The gap is most visible in two specific scenarios. The first is a bridal suite setup shot — FX3 on an 85mm wide for the getting-ready tableau, FX30 on a 35mm insert camera for detail shots. You cut them together in the edit, and the bride's face reads two different temperatures across the same five-minute sequence. Guests never consciously identify it, but it reads as wrong. Something feels off about the footage quality, and the client can't tell you why.
The second is an interview or testimonial rig at a destination wedding — maybe a morning mehendi at a haveli in Jaisalmer, where the morning light is already warm and the subjects are wearing bright pinks and reds. Here the FX30's cyan tendency fights the warmer palette, and the mismatch is amplified by how much red and orange information is already in the scene. The FX3 handles it naturally. The FX30 needs your node.
There is a third scenario worth naming: if you are shooting under LED panel lights tuned to a daylight preset (common at indoor corporate events and some modern wedding halls), the FX30's relative insensitivity to the 600 nm range intersects with LED's characteristic green spike and cyan push to compound the problem. In that environment, your Hue vs. Hue correction may need to go to +8° rather than +6°. Read your scopes. Don't memorize a static value — internalize the diagnosis.
What This Correction Does Not Fix
This single node handles the hue-level mismatch. It does not address three other gaps you may encounter:
Highlight rolloff character. The FX3 has a softer, more cinematic rolloff in the upper stops — a product of its full-frame sensor well depth. The FX30 clips more abruptly. A Luma vs. Saturation curve pulling saturation down as luminance exceeds 75 IRE will help on FX30 clips.
Noise character. The FX3 at ISO 12800 looks fundamentally different from the FX30 at ISO 12800. Noise reduction must be handled per-camera, not globally. DaVinci's Temporal NR at Spatial Threshold 10, Temporal Threshold 15, Luma Mix 0.8 is a reasonable starting point for FX30 footage in mixed-light situations; the FX3 often needs no NR below ISO 6400.
Lens rendering. If your FX3 is on an A-mount or E-mount prime and your FX30 is on a kit zoom, the micro-contrast and chromatic aberration profiles will differ regardless of camera body. That is a lens problem, not a sensor problem, and it lives upstream of any node you build in Resolve.
Correct for those separately, in their own nodes, in the right order. But the skin-tone correction described here is the foundation. Get that right first, and everything else becomes a much shorter conversation.
The FX30 is an extraordinary camera for the price, and on its own terms it produces images that grade beautifully. The issue is never about the camera being inadequate — it is about understanding what its sensor sees differently, and knowing exactly where in your node tree to account for it. One qualifier, one hue rotation, six stops of matte refinement. That is all it takes to make two Sony bodies read as one.