Fix Shot Matching Across Cameras (Corporate Video) | Leumos AI

I've been grading corporate and branded work for four years — Resolve Certified, BFA Cinematography, the whole bit — and the job I dread most isn't the moody indie feature or the music video shot on 16mm. It's the two-day brand shoot where the producer wanted a Sony FX6 main, an A7S III B-cam, iPhone Pro b-roll, and a Mavic 3 drone in the same 90-second hero film. Brand color has to land exactly on the deck spec. The edit drops Friday. It's Wednesday afternoon. And I'm staring at 47 clips that don't match each other in any meaningful way.

If you've ever quoted a corporate package and underestimated how long shot matching would actually take, you already know what I'm about to describe. This piece is a diagnostic — what goes wrong when you mix cameras on branded jobs, why the standard Resolve approach burns half your edit day, and the workflow I've been running on tight turnarounds to keep clients happy without rebuilding node trees from scratch.

Why mixed-camera corporate jobs are uniquely painful

The reason multi-cam matching is hard on corporate work isn't technical — it's contractual. On an indie short, "close enough" can be a creative choice. On a tech brand film for a Series B SaaS company, the marketing director is going to open Photoshop, eyedrop the hero shot, and compare the hex value to their brand guide. If you're off, you redo it.

The cameras don't help. A Sony FX6 in S-Log3/S-Gamut3.Cine has different highlight rolloff than the A7S III running the same gamma — same sensor family, different DSPs, different noise floors. The iPhone Pro in ProRes Log is gorgeous but its color science isn't Sony's; the magentas in skin tones lean differently. DJI's drones in D-Log M are their own animal — neutral on paper, but the gamma curve doesn't sit cleanly under a standard S-Log3 to Rec.709 LUT.

So you end up with four "log" sources that all need different input transforms, different exposure offsets, and different secondary work before they share a believable look. That's before you've even touched the brand grade.

The node-soup problem

Standard Resolve workflow on a 47-clip multi-cam corporate job looks like this:

• CST or input LUT node per camera (four different transforms)
• Exposure match node, eyeballed against a hero shot
• A primaries node for white balance correction
• A secondary qualifier node for skin if there's a presenter
• A shared "look" group with the brand grade
• A final trim node per clip

That's six nodes minimum, per clip, on a timeline where most clips run 3-8 seconds long. You're not grading. You're doing data entry. Matti Haapoja's commercial work for Sony and DJI looks effortless because he has time, a dedicated colorist, and a brand client that lets him push the look. Most of us quoting corporate gigs at $1,200-3,000 a day don't have that runway. We have six hours and a marketing manager who wants to see a v1 by EOD Thursday.

The bottleneck isn't your eye. It's the volume of repetitive setup before the creative work starts.

If you're a corporate video freelancer, we're building this for you. Leumos AI launches in ~30 days — join the early-access list and you'll be in the first 500 (50% off the first year).

A faster workflow for FX6 + A7S III + iPhone + drone

Here's the order I run on multi-cam corporate jobs now. It's the same logic Resolve teaches, just collapsed into fewer steps so I can actually look at the footage instead of building nodes.

Step 1 — normalize the gamma. Every clip gets put into Rec.709 space first. Not a creative LUT, just a transform. The FX6 and A7S III S-Log3 sources go through an S-Log3/S-Gamut3.Cine to Rec.709 transform. The iPhone Pro ProRes Log goes through Apple Log to Rec.709. The drone D-Log M gets its dedicated transform. This is non-negotiable — you cannot match cameras that aren't living in the same display-referred space. The Input Color Space LUT handles this in a click per camera in my current Leumos workflow.

Step 2 — match exposure and white balance to a hero clip. Pick the cleanest A-roll clip with your presenter (or product hero shot) lit properly. That's your reference. Everything else gets pulled toward it. Don't trust the camera's white balance metadata — the FX6 set to 5600K and the A7S III set to 5600K will not match. They're different sensors interpreting the same number.

Step 3 — equalize the timeline as a group. This is where automation saves hours. Instead of touching each clip's exposure/contrast/sat sliders, I let Match All push every clip toward the average baseline of the timeline. It's not perfect — it gets you 70-80% of the way — but it kills the dispersion. Once everything is in the same neighborhood, the creative grade lands consistently. Combine that with AI Scene Cut Detection on the front end and you're not even setting up a node per clip — the upload arrives already chopped into a shot timeline with thumbnails.

Step 4 — drop a reference still for the brand look. This is the step that used to take me the longest. Building a brand look from scratch — finding the right hue rotation, the right shadow density, the right skin protection — is hours of work. Now I'll grab a still from the brand's existing campaign (or a Daniel Schiffer brand film with the energy the client referenced in the brief) and use Reference Image Grading to pull my footage toward it. The intensity slider lets me dial it back to 40-60% so I'm not over-cooking the grade.

Step 5 — surgical fixes on the outliers. There will always be 4-5 clips the automation can't save — a window blew out behind the presenter, the drone shot has a magenta cast from the gradient ND, the iPhone b-roll is two stops under. Manual Primaries for the basics, and if a shot is genuinely broken I'll send it back to the editor and ask for an alternate.

That's it. The six-hour node-soup grade becomes a 90-minute pass, and I'm spending the saved time on the brand-color compliance step — which is the part the client actually checks.

Brand color compliance: where corporate jobs get sunk

Most technical writing on color grading skips this entirely, but on branded jobs it's the only metric that matters at delivery. The brand red has to be the brand red. If the client's primary is Pantone 485 C — a specific warm red used by a lot of consumer brands — you need that to land in your hero shots, on the polo shirt, the product packaging, the wall paint behind the CEO.

AI grading tools (including Leumos) are NOT a substitute for this work. I want to be honest about that. The reference grade gets you to a believable, cohesive look. It does not enforce Pantone compliance. You still need to:

• Pull a frame with the brand color visible
• Use a vectorscope to read where that color is sitting
• Use a HSL qualifier or a hue/sat curve to push it toward the spec value
• Render and check on a calibrated display

What automation buys you is the time to do this step properly instead of skipping it because you ran out of clock.

Frame.io C2C and the 1-3 day turnaround

Half the corporate work I do now lands in my Frame.io inbox before the shoot is even wrapped — the DP uploads proxies from set via the FX6's C2C integration, and I'm pulling clips into my grading session that afternoon. This is where browser-based grading actually matters. I'm not always at my color suite. I'm sometimes on a laptop in a coffee shop reviewing a client's feedback on round two.

Resolve is incredible at home. It's brutal on a 14-inch MacBook Pro at 5pm on a Friday when the agency wants three rounds of revisions before Monday. The reason I built Leumos — and yes, I built it because I was tired of node soup on every gig — is that the browser layer means I can pull a Frame.io proxy, drop a reference, render an h.264 review back to the client, and be done before the espresso cools. That's the real unlock for corporate freelancers on a 1-3 day turnaround.

What AI grading still won't fix

Be honest with your clients about this. AI color matching, including everything Leumos does, struggles with:

• Mixed lighting on skin. Tungsten practicals + daylight window + LED key on the same presenter. The reference grade will average them into something muddy. You'll need to hand-grade those.
• Brand-Pantone enforcement. As covered above — automation gets you to a look, not to a spec.
• Day-for-night. Creative grades that require heavy color theory shifts (warm shadows, cool highlights, crushed blacks) need your taste, not a reference image.
• Footage so under-exposed it's noisy. No grade fixes three stops under on an iPhone in a dim conference room.

Stillmotion's branded documentary work looks the way it does because the lighting on set was correct. Color grading is the last 20% of a job that was 80% done in pre-production. AI doesn't change that math — it just buys you time to focus on the work that actually matters.

Frequently asked questions

What's the fastest way to match Sony FX6 and A7S III footage in the same timeline?

Normalize both to Rec.709 first using an S-Log3/S-Gamut3.Cine transform — don't try to match in log space, you'll fight the gamma curve all day. Then pick a hero clip (usually the cleanest A-roll with your presenter properly lit) and pull everything else toward it. The two sensors are close enough that exposure and white balance offsets are usually small, but they won't auto-match even at the same Kelvin setting. I run a timeline-wide equalization pass after the input transform, then handle outliers with manual primaries. Total time on a 50-clip job: about 30 minutes versus 2-3 hours in raw Resolve.

How do I deal with iPhone ProRes Log clips alongside Sony S-Log3 on a corporate shoot?

The iPhone Pro's Apple Log needs its own input transform — don't try to force it through a Sony LUT, the color science isn't compatible and you'll get green-shifted skin tones. Apple publishes an Apple Log to Rec.709 LUT; use that as your starting point. After both cameras are in Rec.709 space, the iPhone footage will usually need a slight saturation pull-down (it tends to render warmer than the FX6) and the magentas in skin tones often need a small hue rotation. iPhone b-roll generally cuts well next to Sony A-roll once it's been color-managed properly — the resolution holds up surprisingly well for product detail shots.

Can AI color matching enforce brand Pantone colors?

No, and I want to be honest about that. AI matching tools — Leumos, Colourlab, fylm.ai, all of them — work by pulling your footage toward the average color distribution of a reference image. They are not aware of Pantone specs, hex values, or your client's brand guide. To enforce brand color compliance you still need to manually pull frames with the brand color visible, read the vectorscope, and use HSL qualifiers or hue/sat curves to push the color toward the spec value. What AI matching buys you is the time to do this critical step properly instead of skipping it under deadline pressure.

Is Frame.io C2C worth setting up for corporate freelancers?

If you're working on 1-3 day turnarounds for brand clients, yes — it's the single biggest workflow upgrade I've made in the last two years. The FX6's native C2C integration means proxies hit your Frame.io inbox as the camera rolls. You can start your grading session before the DP has even broken down the kit. The downside is bandwidth on remote shoots and the monthly Frame.io cost (usually $20-40/mo per seat depending on plan). For corporate work where the agency expects same-day rough cuts, it pays for itself on the first job. For wedding or event work, it's overkill.

How long should multi-cam matching take on a 90-second corporate film?

Depends on shot count and how disciplined the camera ops were about exposure. For a typical 90-second hero film with 40-60 clips from 3-4 cameras, I budget 60-90 minutes for the matching pass and another 30-60 minutes for brand-color compliance work. In raw Resolve with manual node trees, the same job is 3-4 hours minimum. The time savings come from skipping repetitive node setup — color space transforms, baseline equalization, and reference-based look application can all be automated. The creative judgment calls (which shot is the hero, where the brand color needs to land) still need you in the chair.

What's better for branded content — Resolve or browser-based grading?

Resolve is still the deeper tool — if you're building a fully custom creative grade from scratch with heavy power-window work and tracked secondaries, nothing beats it. For the volume work that pays most corporate freelancers' rent — fast matching, reference-driven brand looks, Frame.io C2C round-trips, MacBook Pro on location — a browser-based tool is honestly faster. I run both. Resolve lives on my color suite tower for the deep work. Leumos lives in the browser for the 70% of jobs where the bottleneck is setup time, not creative depth. Use the tool that matches the job.

Why does my drone footage look different from my A7S III even with the same LUT?

Because D-Log M and S-Log3 are not the same gamma curve, even if both are called "log" in the marketing copy. A standard S-Log3 to Rec.709 LUT applied to D-Log M footage will crush the shadows and lift the highlights incorrectly. DJI publishes a D-Log M to Rec.709 LUT specifically for their drones — use that as your input transform. After that, the drone footage will still typically need a small magenta pull (gradient NDs on the Mavic introduce a slight cast) and a contrast bump to match the cinematic look of the FX6. Treating drone footage as its own camera with its own transform fixes 90% of the matching problem.

---

Leumos AI launches mid-2026. The first 500 early-access signups get 50% off the first year. Join the early-access list →