Mariah, lead editor at MKBHD, was staring at hours of robo taxi footage and a script that was actively working against her. The problem was not a shortage of material. It was the opposite: too much of it, all pulling in different directions, and a first cut that left viewers with the nagging sense that Marquez knew something they did not. What the MKBHD team put together in response is a rare thing: a genuine, inside-the-room account of how professional video editing actually works, from the first rough assembly all the way through color, motion graphics, and sound design. The lesson they kept returning to was deceptively simple. Editing is not software. It is a stack of tiny decisions, and every person in the chain is making them.
When the footage fights the story
Mariah’s first challenge on the robo taxi project was a structural one. She had two cameras, multiple rides, and a script that told her to intercut Marquez’s commentary with the on-site footage. The first cut forced a collision between present-tense talking-head material and past-tense B-roll, and the seams showed. ‘A lot of the times I’m my own worst enemy,’ she said, ‘but sometimes Marquez is my own worst enemy.’ The fix came from sharing the cut with fresh eyes and then making a harder call: removing a genuinely funny scene entirely. Marquez had left his phone behind in one of the autonomous taxis, and the car’s automated system even chimed in to remind him not to forget his valuables. Mariah kept none of it. The scene was real and it was amusing, but it pulled the video away from its actual purpose. What you cut, she noted, is as important as what you keep.
The second project flipped the problem. A video of Marquez explaining how he picks his topics had almost no visual material to work with, which meant the first cut leaned on archive footage and landed flat. Mariah’s solution was counterintuitive: instead of building a flashy montage opener, she leaned into the archive nature of the footage directly, framing a clip from 11 years ago not as a liability but as a memory. ‘A good intro doesn’t just hook the audience,’ she observed, ‘but it also sets up the expectation for the viewer.’ She also added short transitional music passages, following a specific internal rule: Marquez does not like music running for more than about 20 to 30 seconds because it starts to feel like an ad, but with no product to advertise, the music could serve a different function entirely, signaling to the viewer’s brain that a new section was coming.
Color, motion, and the sounds nobody notices
Vin, the art director, took over the grade in DaVinci Resolve and walked through a principle that runs through the whole operation: 80 percent of the image should be done on camera, and the remaining 20 percent is what color grading can actually address. He was skeptical of manufacturer-supplied LUTs, noting that Canon’s CLOG 2 shoots differently across the C70, the R5, and the C300 Mark II, making a single stock conversion unreliable. For Marquez’s A-roll, shot on a Canon C70, he used a custom LUT from a YouTube collaborator whose preset is built on Arri’s color space, then added a plugin called Dehancer at around 400 dollars for two seats, which applies camera-specific film emulation including halation, which adds a slight red flare to highlights and gently warms the image toward a retro quality.
Michael, head of motion design, built his graphics in Adobe Illustrator first before touching After Effects, using vectorized paths so every element could move independently. The three animation principles he returned to were easing, timing, and staging. He used a plugin called Motion Studio, designed by a creator named Matt who runs a YouTube channel called Mount Moaph, to dial in precise easing values on every keyframe rather than wrestling with After Effects’ graph editor by hand.
Ellis handled music and sound design, and his core argument was that music selection is not about personal taste. It is about what already lives in the viewer’s memory. For the OnePlus 13S intro, he needed to make a phone rotating in empty space feel physical and massive. He used large, low-pitched whoosh sounds to match the easing on the phone’s turns, and for one final rotation where the easing had glitched out, he matched the abrupt stop with the sound of a boulder. For the tracing animation at the top of the same intro, rather than using pencil-on-paper sounds, he used a roadside flare and a bell tree, replicating the exact sound signature of the Disney Channel logo reveal that a generation of viewers heard thousands of times as children. Nobody in the comments questioned it.
Rufus and the air conditioner in the room
Rufus, the newest member of the team, joined the conversation not to talk about editing but to make the case that good audio begins before anyone opens a timeline. At MKBHD, the shotgun microphone for Marquez’s A-roll sits just outside the frame to his side, not above him and not below him. Above would point the mic toward the ceiling where the sky panels run their fans. Below would aim it at the desk, which reflects Marquez’s voice back into the mic like shooting him through a mirror. The position that survived was the one that put the microphone’s directional cone on the voice and away from every competing noise source in the room. The target recording level, he noted, should average around minus 20 decibels, with the loudest moments staying around minus 12. His first microphone recommendation for anyone starting out was the Audio Technica AT875R, which he described as having no reason to sound as good as it does relative to its price.
The Audio Technica AT875R sits at one end of a deliberate hierarchy Rufus laid out: starter mic, skip the mid-range entirely, and jump to flagships like the MKH416 or the DPA 4017 only when you are ready to treat audio as a long-term commitment.
Marquez’s own contribution, woven through the finished cut that runs under all of this, was a framework for deciding what videos to make at all. He looks backward to look forward, connecting new topics to older ones that landed well. The robotic hand video made sense partly because a previous video about 3D-printed prosthetics for pets had been well-received, even without huge view counts. And he stopped making top-five countdown videos not because they stopped working but because other channels made them their identity, and chasing that format felt like running on a treadmill: ‘you end up kind of burning out a little bit.’
The Dehancer plugin Vin uses costs about 400 dollars for two seats and performs camera-specific film emulation because, as he put it, the company that made it tests every camera individually rather than applying a single generic conversion.
