Nobody notices the colour grade. Everybody sees it

People usually explain the difference between a film that looks expensive and one that looks cheap in terms of the camera. It is rarely the camera. A good share of the difference is made in a room with no camera in it, after the shoot is over, by people adjusting things that the eventual audience will never be able to point at.

Colour grading is the clearest case, partly because it fails so visibly while staying so hard to name. When a grade is wrong, nobody watching says the grade is wrong. They say the product looks fake, or the room looks flat, or they say nothing and simply do not believe it.

MINI and Mitsubishi are two campaigns we did not shoot. The footage arrived already filmed, and our part was the colour, plus the edit on the MINI 1499 GT. Almost all of that film lives in the dark, which sounds like it would make the job easier and makes it considerably harder: black has to stay black across every cut, and the yellow GT stripe has to be the same yellow in the badge, in the headlight and against the white backdrop. Car paint is about the most difficult surface there is to colour grade, for a reason that has nothing to do with the paint. Paint is a mirror. A mirror is every colour in the room, so you are never grading the car, you are grading whatever was standing around it that day. The Mitsubishi job was five ten second spots that had to play as one set, and that is the same problem spread across five files instead of one.

Grading is the obvious example, but the same thing happens earlier, in decisions about what a film will refuse to do.

The Google Jamboard film has no feature list and no voiceover. It is people at a whiteboard, thinking, in rooms that look like rooms, and what they are planning on the board is the office Christmas party. Everything on that screen was drawn live, which meant redrawing the whole board between takes. We could have composited the drawings in afterwards and saved most of a day. The reason not to is that a faked screen reads as faked to almost everybody, while almost nobody can tell you what gave it away. That is an expensive way to protect something the audience will never consciously register.

Winfrey H. is the same argument outside film. A jewelry house came out of a split needing a name of its own, and the wordmark had to work in a room finished in pale stone and polished steel, standing next to stones worth more than the room. Anything with presence would have been competing with the product. So it is a high-contrast serif, letterspaced wide, no monogram and no ornament, and it had to hold at two sizes with very little in between: cut in metal across the facade, and stamped small on a card. Visitors do not read the sign as a design decision. They read the building as expensive, which was the brief.

None of this is an argument that the shoot does not matter. It is an argument about where briefs go wrong. The things that are easy to write down are shooting days, deliverable counts and resolutions, and those are also the things that are easy to compare between two studios. Colour grade, edit and the long list of small refusals are hard to describe in a line item and nearly impossible to compare on paper, so they tend to get filed under finishing, scheduled last, and cut first when the number comes back too high.

Cut them and you still have the footage. It just looks like footage.