Filming a horror movie in a Hokkaido apartment.

Sapporo was supposed to be the reset. I landed, opened the door of the apartment, and found a room roughly the size of the bed inside it. Japanese apartments are famously small; the ones built for snow country are smaller again, because every cubic metre is a heating bill. Add a week of weather that never lifted, a floor that announced every step, and a door that moved on its own, and the location had already written the film. I just had to admit what it was.

The location wrote the genre

The honest version of that first week is a bad travel vlog: cramped, grey, jet-lagged, nothing photogenic. As a horror set it is excellent. Low ceilings force a wide lens close to the subject, which is the distortion horror uses on purpose. A single window in permanent overcast gives you soft directional light and deep falloff for free. Creaking is production sound most sound designers have to fake.

The reframe cost nothing and changed everything. Same flat, same weather, same week. One is a complaint, the other is a short film.

What actually scares a founder

A ghost-hunting bit is funny for about thirty seconds unless the fear underneath is real. The joke that holds this one together is the answer to what genuinely keeps me up: not the apartment, marketing. Building the thing is the part I know how to do. Making people aware it exists is the haunted corridor.

That gives the film a spine a pure gag does not have. The horror beats track a real anxiety, so the punchline lands as confession rather than skit, and the business update arrives without a single slide.

Music is the tension, everything else is decoration

If you cut the score out of this video it stops working. Not because the images are weak, but because dread is a time-based effect and music is the only tool that controls time directly.

Three moves do most of the lifting. A sustained low drone under dialogue that says nothing is wrong, so the viewer feels wrong before they are told anything. Silence immediately before the beat that matters, because an audience trained on constant sound reads silence as a threat. And sound that arrives before its picture, so the cut answers a question the ear already asked.

Practical note: cut the picture to the music, not the other way around. Laying score over a locked edit gives you a video with music on it. Editing to the track gives you a video that feels composed.

The grade became a product

The look needed a corridor drowned in sick emerald with the practical lights still burning white, which is a harder note than it sounds: flood a scene with green and the lamps go green too, and the frame reads as a filter instead of a place.

Getting there meant building the grade from scratch rather than reaching for a preset. Which raised an obvious question once it worked: this is an asset, why is it sitting on one drive?

So the grades were packaged as .cube files and published free on vanpelt.studio, including the Horror Green and Horror Blue looks from this video, in exchange for an email address. That collected 300 emails in under a week, from an audience of exactly the people the studio is built for: creators who shoot log and grade their own work.

Why that matters more than the view count

The film is good. The distribution move is the part worth copying.

A video is a spike. It gets watched, it decays, and next month you start again. A free asset that solves a real problem is an annuity: it keeps collecting long after the video that produced it stops being recommended. The two reinforce each other, because the LUTs are credible precisely because you can watch the video they came out of.

The video

A spike.

• -Watched hardest in the first week
• -Decays as the algorithm moves on
• -Proves the grade is real
• -Cannot be reused; the next one starts from zero

The free asset

An annuity.

• -Collects long after the video stops being recommended
• -300 emails in under a week, from people who grade their own footage
• -Credible only because the film exists
• -Costs nothing to keep serving

The sequence is repeatable for almost any brand: make the thing, notice which part of the process produced a reusable asset, give that asset away properly, and let the content prove it works. Most companies do the first step, skip the second, and then buy ads to replace the audience they could have kept.

What carries over to client work

Two things travel from this shoot into brand work. First, constraint is a brief. A location that is wrong for the obvious video is usually right for a different one, and the different one is cheaper because you stop fighting the room. Second, a production should leave something behind: a grade, a template, a set of stills, a checklist. Ask at the start what the shoot produces besides the shoot.

See Video for how this gets made for a company, or how to brief a founder story video if the thing you need told is your own.

Sources and further reading

• Free cinematic LUTs for S-Log3, C-Log3 and V-Log

  Vanpelt StudioThe Horror Green and Horror Blue grades from this video, plus the rest of the library. This is the page that collected the emails.

• The free YouTube creator course

  Vanpelt StudioThe larger version of the same idea: give the process away, let the work prove it.

• Write a beat sheet for a YouTube video

  Vanpelt StudioPlanning a short film around tension beats before the location talks you out of it.