Log to Rec.709 Conversion LUTs

Everything else on this site expects an image that has already been converted. These six files are that conversion and nothing else: each one undoes a single camera's log curve. Take the one that matches the format you recorded. There is nothing here to audition.

Reference frame graded with Sony S-Log3 → Rec.709
Preview unavailable
Sony S-Log3 .cube

Sony S-Log3 → Rec.709

Technical S-Log3 / S-Gamut3.Cine to Rec.709 conversion. This is a neutral normalisation, not a creative look — it puts log footage on a correct display-referred starting point so you can grade from there. Built from the published transfer function and gamut matrix, with a highlight shoulder so over-white detail rolls off instead of clipping. 33×33×33.

Reference frame graded with Canon C-Log3 → Rec.709
Preview unavailable
Canon C-Log3 .cube

Canon C-Log3 → Rec.709

Technical C-Log3 / Cinema Gamut to Rec.709 conversion. This is a neutral normalisation, not a creative look — it puts log footage on a correct display-referred starting point so you can grade from there. Built from the published transfer function and gamut matrix, with a highlight shoulder so over-white detail rolls off instead of clipping. 33×33×33.

Reference frame graded with Apple Log → Rec.709
Preview unavailable
Apple Log .cube

Apple Log → Rec.709

Technical Apple Log / Rec.2020, iPhone 15 Pro and later to Rec.709 conversion. This is a neutral normalisation, not a creative look — it puts log footage on a correct display-referred starting point so you can grade from there. Built from the published transfer function and gamut matrix, with a highlight shoulder so over-white detail rolls off instead of clipping. 33×33×33.

Reference frame graded with RED Log3G10 → Rec.709
Preview unavailable
RED Log3G10 .cube

RED Log3G10 → Rec.709

Technical Log3G10 / REDWideGamutRGB to Rec.709 conversion. This is a neutral normalisation, not a creative look — it puts log footage on a correct display-referred starting point so you can grade from there. Built from the published transfer function and gamut matrix, with a highlight shoulder so over-white detail rolls off instead of clipping. 33×33×33.

Reference frame graded with Panasonic V-Log → Rec.709
Preview unavailable
Panasonic V-Log .cube

Panasonic V-Log → Rec.709

Technical V-Log / V-Gamut to Rec.709 conversion. This is a neutral normalisation, not a creative look — it puts log footage on a correct display-referred starting point so you can grade from there. Built from the published transfer function and gamut matrix, with a highlight shoulder so over-white detail rolls off instead of clipping. 33×33×33.

Reference frame graded with ARRI LogC3 → Rec.709
Preview unavailable
ARRI LogC3 .cube

ARRI LogC3 → Rec.709

Technical LogC3 EI 800 / ALEXA Wide Gamut to Rec.709 conversion. This is a neutral normalisation, not a creative look — it puts log footage on a correct display-referred starting point so you can grade from there. Built from the published transfer function and gamut matrix, with a highlight shoulder so over-white detail rolls off instead of clipping. 33×33×33.

Nothing here to audition

The way you pick a look — open a few, see which one you like — is the wrong operation on this shelf. A look is a preference and you can reject it for any reason you have. A conversion is a match: either it is the format your clip was recorded in or it is the wrong file. Liking the picture is not evidence you have the right one.

Six formats are covered: Sony S-Log3, Canon C-Log3, Apple Log, RED Log3G10, Panasonic V-Log and ARRI LogC3. If your camera recorded something else, nothing on this shelf is a near-enough substitute for it.

The measurement can tell these from a look on its own

The profiler that fills the table on every file page reports where a file is steepest, because that is where output moves fastest per unit of input and so where the separation in the picture ends up. 176 of the 179 .cube files in this site's uploads directory that profile as complete lattices are steepest somewhere between 35% and 65% input, read on 23 September 2026: the middle, where an S-curve does its work.

A conversion is steepest at 2.5% input, at the very bottom of the range, and middleware/lutprofile.js prints that reading at all only because these six fall outside the band every look falls inside. So a conversion's page carries a sentence no look's page does, saying it builds its separation low, steepest at 2.5% input. That is the shape of a log curve being undone, not of a look being applied.

Accuracy is the job

These are baked at 33×33×33. Baking the same Sony S-Log3 transform at 17 points instead and measuring both against the analytic answer at 912,673 points gives a mean error of 7.07 code values against 2.75 — one row of the table in what LUT lattice size actually costs, which carries the other seven cameras.

Which slot the file goes into is not the same decision as it is for a look, and both are laid out in installing .cube LUTs in Resolve, Premiere and Final Cut. Once the clip is converted the look goes on top of it, and the rest of the catalogue starts there.

Advertisement

Browse every vibe: Clean & Natural · Cold & Desaturated · Dark & Moody · Film Emulation · Log → Rec.709 Conversion · Neon & Vibrant · Warm & Golden