A LUT (lookup table) is a fixed recipe that maps every input color value to an output color value - apply it, and the same transform happens the same way every time, on any machine, in any software that reads it. That predictability is why LUTs carry looks between departments and why camera vendors publish them for log formats. One piece of folklore is worth correcting up front: the .cube file everyone shares is a de facto standard with its origin at IRIDAS (the company Adobe acquired in 2011), not an official standard from the Academy or ASC - the Academy's own promoted format is CLF, a different, XML-based format. Here is how LUTs actually work, and where they silently change what clients see on a review link.
What a LUT actually does
Every pixel's RGB values go into the table; the mapped values come out. The transform is blind - a LUT knows nothing about exposure intent or scene content, it only knows the map. That is its strength (identical behavior everywhere, no rendering engine required) and its limit (it cannot adapt to the shot, unlike the colorist's wheels). The color pipeline around it - capture space, working space, display space - is in the color-space explainer; a LUT is one bridge between those spaces, and a standard example is the Input Transform that maps a camera's encoding into a working space.
1D vs 3D LUTs
| Type | What it maps | Typical job |
|---|---|---|
| 1D LUT | Each channel independently - three tone curves (R, G, B) with no cross-channel interaction | Gamma, contrast, channel balance adjustments |
| 3D LUT | The full color cube - every input RGB maps to an output RGB, so colors interact (a saturated red can shift differently in shadows than highlights) | Camera-to-display transforms, film emulation looks, creative grades |
DaVinci Resolve's reference manual states the working reality directly: "DaVinci Resolve uses both 1D and 3D LUTs", and documents the 3D lattice sizes that matter - Resolve-created 3D LUTs are 33×33×33 cubes with 32-bit floating-point processing, and exports can be 17³, 33³, or 65³. The lattice is an approximation of a continuous transform: finer cubes track subtle color separation better, at the cost of file size. The manual also notes a 33×33×33 cube holds around 35,937 color samples - the grid the transform is interpolated across.
The .cube format, honestly
The format that made LUTs portable is .cube, created by IRIDAS for its Colorista/Framecaster tooling before Adobe acquired the company in 2011. It spread because every vendor adopted it, not because a standards body wrote it - there is no formal published spec to cite, and the Academy's answer to that gap is CLF (Common LUT Format), an XML-based format. Resolve's own manual carries the positioning: CLFs are "promoted by the academy as the ideal LUT format for use with ACES," and ACES look transforms are recommended in CLF for its precision. Practical rule: .cube for everyday interchange (it opens everywhere), CLF where ACES precision matters (what ACES is, explained here).
Technical vs creative LUTs
The split everyone uses in speech - technical LUTs that convert between spaces (log-to-Rec.709 conversion LUTs, display calibration LUTs) versus creative LUTs that impose a look (film stocks, brand grades) - is industry convention rather than a standards-body definition, and it maps cleanly onto the pipeline: technical transforms sit at the input and output ends, creative transforms sit in the middle where grading decisions live. The failure mode is order: apply a creative look LUT to footage that has not been converted out of log, and the look lands on the wrong encoding - the "flat, wrong, muddy" result that gets diagnosed as a bad LUT when it is actually a sequence error. Camera vendors publish conversion LUTs for exactly this reason (the review-side habit of applying a conversion LUT before anything else is why ungraded drone or log footage should never be sent to clients - the drone review workflow builds that in).
Where LUTs meet the review link
Two review-side surprises trace back to LUTs. First, the look that never made it: a review export rendered without the creative LUT baked in shows clients a flat log image - the diagnosis list for that is in why the review link looks different, and the fix is exporting the review copy with the look applied. Second, the double transform: a conversion LUT applied twice (in the timeline and in the export) crushes contrast and oversaturates - check which layer owns the transform before exporting either the review copy or the master.
FAQ
Is .cube an official industry standard?
No - it is a de facto one. The format originated at IRIDAS (acquired by Adobe in 2011) and spread through universal vendor adoption; no standards body published its spec. The Academy's standardized answer is CLF, an XML format promoted as the ideal LUT format for ACES work.
Do LUTs reduce image quality?
A well-built 3D LUT at 33³ or 65³ lattice size with floating-point processing is effectively transparent for normal grading work; quality loss appears at coarse lattice sizes, extreme input values outside the table's range, or from stacking transforms that were never meant to combine.
What is the difference between a technical and a creative LUT?
Industry convention, not a formal spec: technical LUTs convert between color spaces (log-to-Rec.709, display calibration); creative LUTs impose a look. Technical ones belong at the pipeline's input/output ends, creative ones in the middle - and the order matters, because a look applied before the conversion lands on the wrong encoding.
Why does my log footage look flat without a LUT?
Because log is a recording encoding, not a display encoding - it keeps highlight and shadow data in a flat curve designed for grading headroom. A conversion LUT maps it to display space for viewing; without one, the flat image is the data being honest about what it is.
My graded export lost its look on the review link - is that a LUT problem?
The most common cause is a review copy exported without the creative LUT baked in, showing clients the pre-look image. Re-export the review copy with the look applied, and if it still looks off, work through the five sources in the review-link color-shift diagnosis.