Colour Grade Shows Whether Anyone Supervised the Edit
You are watching an agency reel with the sound off, and every clip looks fine alone. Did anyone look at the whole thing before it went out? Colour grade is the quickest way to find out.
Why grade is the cheapest check on a reel
Grade is the colour and tonal treatment applied after the footage is cut, separate from fixing exposure on a single clip. Correction makes clips look natural and match each other, while grading gives them a deliberate look. The terms are often used interchangeably, as the reference article on grading notes.
It needs no metrics or inside knowledge, only a screen and patience. Finishing comes last, so unsupervised work tends to give itself away there. A reel built from several shoots is the hardest place to hide the absence of a finishing owner, so it is the best place to look.
This piece covers grade only. For cuts, read the editing decisions that separate good social video from filler. For graphic identity, see whether an agency built a system or just one good post.
Inside one video: the cut-to-cut test
Pause on the frame before a cut, then the frame after. Compare faces and whites. Look for three faults.
The first is skin tone drift. The same person, or a similar complexion, goes warmer, greener or paler across a cut with no change in the scene.
The second is the white point wandering. A white shirt, wall or page is cream in one shot and faintly blue in the next. Your eye forgives this at playback speed.
The third is a stacked preset on mixed lighting. A LUT is a lookup table that swaps each input colour value for a fixed output value, so one preset applies the same transform to every clip whatever the light was like on the day. Lay it over daylight, tungsten and ring light footage and the results split. Some shots look rich, others sick, with greens and skin turning muddy where the light did not match what the preset expected.
The acceptable version is not identical footage. Shots from different lighting still sit within one range of contrast and colour temperature, though never the same at capture.
Blacks, highlights and the phone screen
Tonal faults show most on a phone. Look for these:
- Crushed blacks. Dark areas collapse into flat black, losing detail in hair, a dark jacket or a night shot. Dramatic on a bright monitor, a hole on a phone at low brightness.
- Milky or lifted blacks. The opposite fault: a hazy grey wash, often a flat clip never graded at all.
- Clipped highlights. Windows, skies and white products blow out into a flat patch.
- Brightness dependence. Watch on a phone at normal brightness, then with it turned up. Be suspicious of work that only survives one of the two. See judging social work at the size it actually shipped at.
- Honesty about screens. Screens differ, so one odd frame proves nothing. The signal is a pattern across many shots.
The mixed-footage reel: designed or assembled
Picture a reel of client shoots, phone clips, stock and animation. In the assembled version, each client section carries its own colour: warm and soft, then cool and hard, then flat and grey. The reel reads as a stack of separate videos with a logo animation between them.
In the designed version, the sections share a contrast curve, a black level and a way of treating skin. Skin sits in the same place in an office, a street or a kitchen, and the differences that remain look chosen. A consistent grade does not mean a single look, and different clients can have different palettes. The test is whether each palette is held within its section and the move to the next looks decided.
Stock is the usual giveaway. Was it brought into the same finish or left looking like a different film? Telling a custom shoot from treated stock footage goes further.
To run the check, pick three shots from different parts of the reel, freeze them side by side, and ask whether they could plausibly share one cutting room. If you hesitate, you have your answer.
Across a month of posts, not just one video
A live account shows what one reel cannot. Scroll the last month of video and look at frames in a row. A held grade is a recognisable colour temperature and contrast level that survives different locations and people.
Then separate drift from deliberate change. A seasonal shift or new campaign look is a decision and usually arrives with a clean break. Slow, unannounced drift, or one post that suddenly looks different for no reason, points to nobody owning the finish. A grade can also be consistent and bad: consistency shows supervision, not taste.
What a grade cannot tell you
| Limit | What to do about it |
|---|---|
| Rough or documentary work may reject a polished grade on purpose. | Ask whether the rawness is consistent and chosen, not whether the footage is pretty. |
| A uniform grade can cover a weak idea. | Grade is one craft signal among several. It does not replace judging the idea, the edit or the strategy. |
| Upload recompression and filters can shift colour and blacks. | With a recompressed or flattened copy, treat small differences as unreliable. See the post on work you can only see flattened. |
Keep this:
- Faces across cuts.
- Blacks on a phone.
- Three frozen frames from different parts of the reel.
See the selections
Our selections are editorial picks. See this year’s selections, or nominate an agency whose finishing holds up.
FAQ
How can I tell if a social video was colour graded properly?
Pause either side of a cut and compare skin and whites, then check that dark areas keep detail on a phone.
What is the difference between colour correction and colour grading?
Correction makes clips look natural and match each other, and grading adds a deliberate look on top, though many people use the two words interchangeably.
Does every video on an account need the same grade?
No. Clients and campaigns can differ. What matters is that each look is held and changes look chosen.
Can I judge colour on a phone?
Yes, and you should, since that is where most of it is watched. A pattern counts; one odd frame does not.
Sources
- Color grading, Wikipedia, fetched 2026-10-13
- Look-up table, Wikipedia, fetched 2026-10-13