Free download

Simple LUT Tester

A single PNG that tells you whether a LUT puts colour where it belongs — or quietly moves it somewhere else. Works with any camera, any log format.

Drop it on your timeline, apply the LUT, look at your vectorscope. Six corners on their marks means the LUT is placing colour correctly. Corners flung off means it isn't, and no amount of grading will put that back.

Simple LUT Tester 16-bit PNG · 1920×1080
The test chart: two hexagonal views of the RGB colour cube, a row of colour bars, and a grey ramp with a percentage scale beneath it.
Two views of the RGB cube, 75% bars, and a grey ramp with a code-value scale.

How to use it

Four steps, then one thing to look at.

  1. Put the PNG on your timeline.
  2. Apply the LUT you want to test.
  3. Open your vectorscope.
  4. Check where the six corners land.

That is the whole method. There is nothing to configure for your camera — the same file works whether you shoot S-Log3, D-Log, Apple Log, V-Log or anything else. Why that works is at the bottom if you are curious, and skippable if you are not.

The good, the bad and the ugly

Three real LUTs, the same chart, the same vectorscope. You are looking at the RGB colour cube from one corner — three faces meeting in the middle. Once you see it as a box rather than a hexagon, the difference between these is not subtle.

The good
Vectorscope showing an even, square cube trace with all six corners sitting inside their graticule boxes.

Square and true. Six corners in their boxes, faces even, mesh consistent from centre to edge. Every saturated colour lands where it should.

The bad
Vectorscope showing a cube trace bent out of square, with the magenta corner pushed well outside its graticule box.

Bent out of square. Still recognisably a cube, but magenta is shoved past its box and the faces bulge. Saturated colour is being pushed somewhere it was never meant to go.

The ugly
Vectorscope showing a collapsed, fanned trace with no cube structure left and no corner near its graticule box.

Melted. The cube has collapsed into a fan. Not one corner is near its mark. Nothing in a grade will put this back together.

Reading the shape

  • On the marksThe LUT is placing colour where it should.
  • RotatedIt is shifting hue across the board.
  • Falls shortIt is desaturating.
  • Corner flattenedIt is clipping that colour.
  • Torn, in ribbonsBadly built. This one is unmistakable.
  • Middle off centreYour greys have a colour cast.

A look is allowed to shift hue and saturation a little — that is what a look is. What it is not allowed to do is lose the plot entirely, and that reads at a glance once you know where the corners belong.

What not to read into it

Large flat areas of colour in the hexagons are not a fault. This is worth saying plainly, because it is the obvious thing to assume and it is wrong.

The chart holds the whole range of code values your camera file can contain, including several stops of highlight above diffuse white. Any LUT converting log to Rec.709 has to crush that range into the top few percent of the display — there is nowhere else for it to go. Flat regions are the correct result of that.

Five ways of measuring banding were tested against a correct conversion and a deliberately broken LUT. None separated them, and two ranked the correct one as worse:

Five attempts to catch banding by measurement
TestCorrect conversionBroken LUT
Share of hexagons going flat20–30%32–42%
Distinct output colours45%48%
Tearing under a hard push+25.4 pts+17.5 pts
Measured inside the cube~30% at every depth—
Local gain, slope, resolutionall three tripall three trip

The reason is the same every time. A correct log-to-display conversion has to destroy information — seventeen stops squeezed into six. So anything measuring information loss flags it. That is not a measurement waiting to be fixed; it is what the transform is.

Which is why this page tells you to judge placement and nothing else. Judging banding by eye here condemns good LUTs.

Pushing it further

Some faults only appear once you lean on them.

  • Raise the gammaHard, and after the LUT. Problems buried in the shadows crawl out.
  • Split the channelsLook at red, green and blue on their own. Unevenness in the saturated areas shows there first.
  • Under and overGrade the shot a couple of stops either way. A look built by eye on one correctly exposed frame falls apart here.

That last one is the honest test of a LUT you have paid for. A look that only holds together at the exposure it was built at is a look you cannot use.

Why one chart works for every camera

The chart holds code values, not colours. The corner of the hexagon is simply the largest red number the file can contain, and that number is the same whether you call the file S-Log3, D-Log or Cineon. The label changes; the number does not. Any LUT handling out-of-range colour the normal way lands it on the 100% mark — measured at 0.0 degrees of hue error across conversions built for entirely different gamuts.

An earlier version of this chart did the opposite: it worked out what Rec.709 red ought to be in each camera's colour space. That read wrong whenever the assumption about the camera missed what the LUT expected, and it disagreed with a properly calibrated monitor. The monitor was right.

Built the Simple LUT Tester because I got tired of watching other people's LUTs fail this test. If you want a set that passes it — per camera, per ND brand, X-Rite verified — see the Simple LUTS packs.