Free tool · nothing is uploaded
LUT strength mixer
Turn a LUT down to 40% — or push it past 100% — and bake the result into the file, so the strength travels with it instead of living in your project.
Drop a .cube LUT here
Nothing is uploaded — the LUT is read and rewritten on your own device.
Opacity lives in the project. Strength lives in the file.
Almost every grading application lets you dial back a LUT — node opacity in DaVinci Resolve, effect opacity in Premiere, the mix slider in Final Cut. All of them store that number in the project, not in the .cube. The LUT on disk is unchanged, so the moment it leaves that timeline the reduction is gone.
That matters more than it sounds. A LUT loaded onto a camera monitor, into OBS for a stream, onto a phone app, or handed to a colleague has no opacity control in the chain at all. If the look needs to be gentler in those places, the only way to get it is to write the gentler version down.
What 50% actually means
A 3D LUT is a table: for this input colour, output that one. Every entry is a pair of points in colour space, and the LUT is the instruction to move from the first to the second. Strength is simply how far along that line you stop.
At 0% nothing moves and the table becomes the identity transform. At 100% every colour lands exactly where the LUT specified. At 50% each one travels half the distance — which is not the same as fading the graded image over the original, though the two agree closely for most looks, and differ where the LUT does something non-monotonic.
Frequently asked questions
Why not just lower the node opacity in Resolve?
Do, if the LUT only ever runs in that project. Opacity is a property of the node, not the file, so it does not travel: send the LUT to a collaborator, load it into OBS, or push it to a camera monitor and you get it back at full strength. Baking the reduction into the file makes the weaker version the LUT itself.
How is the strength calculated?
Each of the LUT's entries is moved along the straight line between where that colour started and where the LUT sends it. At 0% every entry sits at its input value, which is the identity transform; at 100% it sits exactly where the LUT put it. 40% means each colour travels 40% of that distance. This is the same interpolation a grading application performs at playback, done once and written down.
Does reducing strength lose quality?
No, and it often gains some. The output grid is the same size as the input, and blending toward the identity moves values closer together rather than further apart, so quantisation error shrinks rather than grows. A 50% LUT is a smoother transform than the original.
What does going above 100% do?
It extrapolates: colours continue past where the LUT put them, along the same line. This is genuinely useful for a look that is too timid, but it is not what the author designed, and channels clamp at 0 and 1 rather than wrapping. In practice that means highlights and saturated colours flatten first. The preview is the check.
Can I do this to a .3dl or a Lightroom preset?
Convert first. The 3DL ↔ CUBE tool brings an Autodesk .3dl across, and CUBE to XMP goes the other way to a Lightroom preset. A preset does not need this tool at all — its sliders are already editable.
Is my LUT uploaded?
No. The file is read with the browser's File API, the blend runs on your CPU and the preview renders on your own GPU. Nothing is transmitted.