Color management, LUTs & HDR mastering
Set color-managed pipelines, apply and build LUTs, and master HDR deliverables to spec.
20 project-authored proposals · taxonomy v1.1 · 0 new benchmark runs. Verifiers and partial-credit rules below are designs, not executed results.
- Typical applications
- DaVinci Resolve · Premiere Pro · Nuke
- Class GUI-need label
HY· Hybrid
Task proposals · 20
Expand to inspect the complete original wording and proposed checks. Original proposal text remains English in both interfaces.
R06.04.02.06-T01 Master an HDR10 deliverable with correct static metadata
Original instruction
In DaVinci Resolve project input/doc_ep3.drp (already DaVinci Wide Gamut color managed), set output to Rec.2100 ST2084 with 1000-nit timeline luminance, keep MaxCLL ≤ 1000 and MaxFALL ≤ 400 by adjusting the grade, and render an HDR10 ProRes 4444 XQ master to out/ep3_hdr10.mov with mastering display P3-D65, 1000/0.0001 nits.
- Platforms
mac· macOSwin· Windowslin· Linux- Applications
- DaVinci Resolve
- Interfaces
desk· Desktop GUI- GUI need · task setting
G2· GUI-essential- GUI need · ordinary access
G2· GUI-essential- Interaction primitives
click· Point & clickvis· Visual judgmentav· Audio/video over time- Work product
WP05· Time-based media- Conditions
- None declared
- Proposed difficulty
- 4 / 5
Proposed verifier
ffprobe/mediainfo: BT.2020 primaries, PQ transfer, mastering display and content light metadata exact; grader HDR analyzer: MaxCLL ≤1000, MaxFALL ≤400; .drp output space and timeline luminance set. External scripting set to None; fusionscript connections detected -> 0; ffmpeg absent from the VM.
Proposed partial credit
30% project output color space and luminance; 30% light levels within limits; 40% HDR10 metadata and codec correct
R06.04.02.06-T02 Bake a node graph into a 3D LUT and verify it
Original instruction
In DaVinci Resolve, timeline 'Promo' clip 1 has a 4-node grade. Generate a 65-point 3D LUT from it (Generate LUT), save out/promo_65.cube, then apply that LUT alone to clip 2 so it matches the graded look of clip 1 on the same source frame.
- Platforms
mac· macOSwin· Windowslin· Linux- Applications
- DaVinci Resolve
- Interfaces
desk· Desktop GUI- GUI need · task setting
G2· GUI-essential- GUI need · ordinary access
G2· GUI-essential- Interaction primitives
click· Point & clickvis· Visual judgment- Work product
WP05· Time-based media- Conditions
- None declared
- Proposed difficulty
- 3 / 5
Proposed verifier
LUT header size 65; grader applies it via OCIO to clip 1's source frame and compares to Resolve's grade render: ΔE2000 mean <= 1.0; clip 2 node graph has only the LUT. LUT file must be written by Resolve (provenance); external scripting disabled and fusionscript use detected -> 0.
Proposed partial credit
50% LUT reproduces grade; 20% 65-point size; 30% clip 2 uses LUT only
R06.04.02.06-T03 Apply a camera LUT to B-cam clips only in Premiere
Original instruction
In Premiere Pro project input/interview.prproj, apply input/Bcam_SLog3_to_709.cube as the Lumetri Input LUT to every clip from camera B (per input/clip_list.csv, 23 clips across 3 sequences), leaving A-cam clips untouched, then export the main sequence as ProRes 422 HQ Rec.709.
- Platforms
mac· macOSwin· Windows- Applications
- Premiere Pro
- Interfaces
desk· Desktop GUI- GUI need · task setting
G2· GUI-essential- GUI need · ordinary access
G2· GUI-essential- Interaction primitives
click· Point & clicknav· Pan/zoom/scrolldrag· Drag & drop- Work product
WP05· Time-based media- Conditions
licensed· Licensed pro app- Proposed difficulty
- 3 / 5
Proposed verifier
Grader parses the saved .prproj XML: Lumetri effect with that input LUT on exactly the 23 B-cam clip instances; export frames at 6 timecodes ΔE2000 <= 2 vs reference; tags BT.709. ExtendScript/UXP panels disabled (CEP debug off, scripts folder read-only); .prproj provenance checked.
Proposed partial credit
50% LUT on exactly the B-cam clips; 20% A-cam untouched; 30% export correct
R06.04.02.06-T04 HDR10 PQ master with MaxCLL/MaxFALL and an SDR trim
Original instruction
In DaVinci Resolve Studio, master timeline 'Feature_R1' as HDR10: Rec.2020 ST.2084 1000-nit output, grade highlights so MaxCLL <= 1000 and MaxFALL <= 400 nits, set mastering display metadata (P3-D65, 1000/0.0001), render HEVC Main10, and render an SDR Rec.709 version via the HDR-to-SDR output transform.
- Platforms
mac· macOSwin· Windows- Applications
- DaVinci Resolve Studio
- Interfaces
desk· Desktop GUI- GUI need · task setting
G2· GUI-essential- GUI need · ordinary access
G2· GUI-essential- Interaction primitives
click· Point & clickvis· Visual judgmentdrag· Drag & drop- Work product
WP05· Time-based media- Conditions
licensed· Licensed pro app- Proposed difficulty
- 5 / 5
Proposed verifier
ffprobe: HEVC Main10, smpte2084, bt2020; SEI mastering display and content light level present; grader-measured MaxCLL <= 1000 and MaxFALL <= 400 and within 5 pct of signaled values; SDR file BT.709 with no clipped skin (ΔE vs reference <= 4). External scripting disabled; fusionscript connections detected -> 0.
Proposed partial credit
30% HDR10 encoding and tags; 30% light-level limits met and signaled; 20% mastering metadata; 20% SDR version
R06.04.02.06-T05 ACES pipeline in Nuke for mixed-camera plates
Original instruction
In Nuke with the ACES 1.3 OCIO config, read plates A001 (ARRI LogC4/AWG4) and B002 (S-Log3/S-Gamut3.Cine) with correct input colorspaces, write ACEScg linear EXRs (PIZ, half) for VFX, and write review MOVs through the Rec.709 ODT. Save comp_prep.nk.
- Platforms
mac· macOSwin· Windowslin· Linux- Applications
- Nuke
- Interfaces
desk· Desktop GUI- GUI need · task setting
HY· Hybrid- GUI need · ordinary access
CL· CLI-sufficient- Interaction primitives
click· Point & clickcode· Code & commands- Work product
WP05· Time-based media- Conditions
licensed· Licensed pro app- Proposed difficulty
- 4 / 5
Proposed verifier
EXR headers chromaticities = AP1, compression PIZ, half; pixel values vs grader reference conversion (OCIO ociocolor) max abs error <= 0.002; MOVs tagged BT.709 and match reference within ΔE2000 1.5; .nk Read/Write colorspace knobs exact.
Proposed partial credit
35% input colorspaces correct; 35% ACEScg EXRs exact; 30% review MOVs
R06.04.02.06-T06 Apply a LUT with ffmpeg and tag color metadata (CLI control)
Original instruction
Apply input/show.cube to input/clip_log.mov with ffmpeg's lut3d filter (tetrahedral), encode ProRes 422 HQ, and tag the output correctly as BT.709 primaries, transfer and matrix in limited (TV) range. Write the command to out/apply.sh.
- Platforms
lin· Linuxmac· macOSwin· Windows- Applications
- ffmpeg
- Interfaces
term· Terminal- GUI need · task setting
CL· CLI-sufficient- GUI need · ordinary access
CL· CLI-sufficient- Interaction primitives
code· Code & commands- Work product
WP05· Time-based media- Conditions
- None declared
- Proposed difficulty
- 2 / 5
Proposed verifier
ffprobe: color_primaries/color_trc/colorspace = bt709, color_range tv, prores profile 3; frames vs grader's OCIO application of the LUT ΔE2000 mean <= 0.8 (detects range or interpolation errors); apply.sh reproduces it.
Proposed partial credit
40% LUT applied correctly; 35% color tags and range; 25% reproducible command
R06.04.02.06-T07 Bake a LogC-to-Rec.709 LUT with ociobakelut
Original instruction
Using the studio OCIO config in input/config.ocio, bake a 65-point .cube converting ARRI LogC4 to Rec.709 Gamma 2.4 through the 'Show Look' look, plus a 1D shaper if needed for accuracy. Save out/logc4_show_709.cube and report max ΔE2000 vs the direct OCIO transform in out/report.txt.
- Platforms
lin· Linuxmac· macOS- Applications
- ociobakelut · OpenColorIO
- Interfaces
term· Terminal- GUI need · task setting
CL· CLI-sufficient- GUI need · ordinary access
CL· CLI-sufficient- Interaction primitives
code· Code & commands- Work product
WP05· Time-based media- Conditions
- None declared
- Proposed difficulty
- 3 / 5
Proposed verifier
LUT size 65 (and shaper if present); grader evaluates on 10k random LogC4 samples: max ΔE2000 <= 1.0 vs direct OCIO processor with the look; report value within 0.2 of grader's measurement.
Proposed partial credit
25% LUT format and size; 50% accuracy threshold met; 25% report correct
R06.04.02.06-T08 HLG broadcast deliverable from a PQ master
Original instruction
Convert the PQ 1000-nit master input/doc_pq.mov to an HLG Rec.2100 broadcast file in DaVinci Resolve (color space transform with tone mapping, narrow range, peak 1000 nits), then render XAVC Intra or ProRes 422 HQ with correct HLG tags.
- Platforms
mac· macOSwin· Windowslin· Linux- Applications
- DaVinci Resolve
- Interfaces
desk· Desktop GUI- GUI need · task setting
G2· GUI-essential- GUI need · ordinary access
CL· CLI-sufficient- Interaction primitives
click· Point & clickvis· Visual judgment- Work product
WP05· Time-based media- Conditions
- None declared
- Proposed difficulty
- 4 / 5
Proposed verifier
ffprobe: arib-std-b67 transfer, bt2020 primaries, tv range; grader decodes and converts back to PQ: per-frame mean nit error <= 3 pct and no clipping above 1000; skin patch ΔE2000 <= 3 vs reference. External scripting disabled; fusionscript connections detected -> 0.
Proposed partial credit
35% HLG tags and range; 45% conversion accuracy; 20% no clipping
R06.04.02.06-T09 Legalize a timeline for broadcast QC
Original instruction
The broadcaster's QC (EBU R103) failed input/spot_30s.mov: luma excursions above 940 and below 64 and out-of-gamut RGB. In Resolve, legalize it with soft clipping and grade tweaks so every frame stays within 64-940 luma (10-bit) and RGB within the R103 preferred range, without visibly changing the grade, and render out/spot_legal.mov.
- Platforms
mac· macOSwin· Windowslin· Linux- Applications
- DaVinci Resolve
- Interfaces
desk· Desktop GUI- GUI need · task setting
G2· GUI-essential- GUI need · ordinary access
G2· GUI-essential- Interaction primitives
click· Point & clickvis· Visual judgment- Work product
WP05· Time-based media- Conditions
- None declared
- Proposed difficulty
- 3 / 5
Proposed verifier
Grader scans all frames: 0 pixels outside 64-940 luma and R103 RGB limits beyond the 1 pct area allowance; mean ΔE2000 vs original on in-range pixels <= 1.5. External scripting disabled; fusionscript use detected -> 0.
Proposed partial credit
45% luma legal; 30% RGB gamut legal; 25% grade preserved
R06.04.02.06-T10 HDR PQ sequence and HEVC export in Premiere Pro
Original instruction
In Premiere Pro create an HDR sequence (Rec.2100 PQ working space) from input/hdr_clips/, set each clip's color space override per input/clip_meta.csv, and export HEVC 10-bit Main10 with HDR10 metadata (MaxCLL 800, MaxFALL 300) to out/promo_hdr.mp4.
- Platforms
mac· macOSwin· Windows- Applications
- Premiere Pro
- Interfaces
desk· Desktop GUI- GUI need · task setting
G2· GUI-essential- GUI need · ordinary access
CL· CLI-sufficient- Interaction primitives
click· Point & clicknav· Pan/zoom/scroll- Work product
WP05· Time-based media- Conditions
licensed· Licensed pro app- Proposed difficulty
- 4 / 5
Proposed verifier
ffprobe: HEVC Main10, smpte2084, bt2020, SEI light level 800/300; .prproj sequence working space Rec.2100 PQ and 6 clip overrides exact; frames vs reference ΔE ITP <= 5. ExtendScript/UXP disabled; .prproj provenance checked.
Proposed partial credit
35% sequence working space and overrides; 40% export encoding and tags; 25% light-level metadata
R06.04.02.06-T11 Dolby Vision analysis and 100-nit trims for three shots
Original instruction
In DaVinci Resolve Studio, enable Dolby Vision (CM v4.0) on timeline 'Trailer_HDR', run analysis on all shots, add 100-nit target trims (lift/gain/saturation) for shots 4, 9 and 15 to match input/sdr_ref_frames/, and export the Dolby Vision XML to out/trailer_dv.xml.
- Platforms
mac· macOSwin· Windows- Applications
- DaVinci Resolve Studio
- Interfaces
desk· Desktop GUI- GUI need · task setting
G2· GUI-essential- GUI need · ordinary access
G2· GUI-essential- Interaction primitives
click· Point & clickvis· Visual judgmentdrag· Drag & drop- Work product
WP05· Time-based media- Conditions
licensed· Licensed pro app- Proposed difficulty
- 5 / 5
Proposed verifier
XML parsed: CM v4.0, L1 metadata for all 22 shots, L2/L8 trims for 100-nit target on exactly shots 4, 9, 15; grader applies the DV mapping via reference tools and the 3 frames match SDR refs ΔE2000 <= 3. External scripting disabled; fusionscript connections detected -> 0.
Proposed partial credit
35% analysis on all shots; 40% trims for the three shots match refs; 25% valid XML export
R06.04.02.06-T12 Fix mis-tagged clip color spaces in a managed project
Original instruction
In Resolve project input/music_vid.drp (color managed), 7 clips shot Rec.2020/HLG were imported as Rec.709 and look washed out. Identify them by eye and scopes, set their correct input color space and gamma, and render the 7 marked timecodes as stills out/fix_<tc>.tif.
- Platforms
mac· macOSwin· Windowslin· Linux- Applications
- DaVinci Resolve
- Interfaces
desk· Desktop GUI- GUI need · task setting
G2· GUI-essential- GUI need · ordinary access
G2· GUI-essential- Interaction primitives
click· Point & clickvis· Visual judgmentnav· Pan/zoom/scroll- Work product
WP05· Time-based media- Conditions
- None declared
- Proposed difficulty
- 3 / 5
Proposed verifier
Project export: exactly the 7 clips have Rec.2020/HLG input, others unchanged; stills ΔE2000 mean <= 2 vs reference. Clip IDs are not listed anywhere (perceptual identification). External scripting disabled; fusionscript connections detected -> 0.
Proposed partial credit
50% correct clips identified and retagged; 15% no false changes; 35% stills match
R06.04.02.06-T13 Export ASC CDLs for VFX pulls
Original instruction
For the 12 VFX shots in Resolve timeline 'Reel2' (marked with blue markers), export per-shot ASC CDL values (slope, offset, power, saturation) of the primary grade as a .ccc collection with ids equal to shot names, and an EDL with the CDL comments, to out/reel2_vfx.ccc and out/reel2_vfx.edl.
- Platforms
mac· macOSwin· Windowslin· Linux- Applications
- DaVinci Resolve
- Interfaces
desk· Desktop GUI- GUI need · task setting
HY· Hybrid- GUI need · ordinary access
CL· CLI-sufficient- Interaction primitives
click· Point & click- Work product
WP05· Time-based media- Conditions
- None declared
- Proposed difficulty
- 3 / 5
Proposed verifier
.ccc XML has 12 ColorCorrection ids exactly matching shot names; SOP/Sat values equal Resolve's node values to 4 dp; EDL events with *ASC_SOP/*ASC_SAT comments for the 12 shots with correct source TC.
Proposed partial credit
40% all 12 CDLs present with ids; 40% values exact; 20% EDL with CDL comments
R06.04.02.06-T14 ACEScct grade to a P3-D65 theatrical deliverable
Original instruction
Set Resolve project input/short_film.drp to ACEScct (ACES 1.3) with IDTs per camera, ODT P3-D65 (48 nits) for theatre, grade shot 1 to input/hero_still.exr, then render DCDM-style 16-bit TIFF XYZ via the ACES output transform for frames 1-48 to out/dcdm/.
- Platforms
mac· macOSwin· Windowslin· Linux- Applications
- DaVinci Resolve
- Interfaces
desk· Desktop GUI- GUI need · task setting
G2· GUI-essential- GUI need · ordinary access
G2· GUI-essential- Interaction primitives
click· Point & clickvis· Visual judgmentdrag· Drag & drop- Work product
WP05· Time-based media- Conditions
- None declared
- Proposed difficulty
- 5 / 5
Proposed verifier
Project settings ACEScct, ACES 1.3, ODT P3-D65 48 nits, per-clip IDTs correct; TIFFs 16-bit XYZ with gamma 2.6 encoding; shot 1 matches hero still ΔE2000 mean <= 2.5 after reference XYZ->P3 conversion. External scripting disabled; fusionscript connections detected -> 0.
Proposed partial credit
30% ACES settings and IDTs; 30% XYZ TIFF output spec; 40% hero match
R06.04.02.06-T15 Fix Mac 'washed-out' exports with the right tags
Original instruction
The client says the Resolve export input/ad_v3.mov looks washed out in QuickTime on Mac. Re-render it from project input/ad.drp so it is tagged Rec.709 (1-1-1) for broadcast deliverable and also deliver a web version tagged Rec.709-A for Apple playback, both from the same grade.
- Platforms
mac· macOSwin· Windows- Applications
- DaVinci Resolve
- Interfaces
desk· Desktop GUI- GUI need · task setting
G2· GUI-essential- GUI need · ordinary access
G2· GUI-essential- Interaction primitives
click· Point & click- Work product
WP05· Time-based media- Conditions
- None declared
- Proposed difficulty
- 2 / 5
Proposed verifier
ffprobe/colr atom: broadcast file primaries/transfer/matrix 1-1-1; web file 1-1-1 with Rec.709-A gamma tag (colr nclx transfer 1 plus gama/Apple tag); pixel data identical between files (PSNR > 50 dB). External scripting disabled; fusionscript connections detected -> 0.
Proposed partial credit
45% broadcast tags exact; 40% web tags exact; 15% same pixel data
R06.04.02.06-T16 Resample a LUT and convert it for Avid (CLI control)
Original instruction
Convert input/look_17.cube (17-point) to a 33-point .cube and to a 33-point .3dl (12-bit integer) for an Avid workstation, using OCIO or colour-science with tetrahedral interpolation. Save out/look_33.cube and out/look_33.3dl.
- Platforms
lin· Linuxmac· macOSwin· Windows- Applications
- OpenColorIO · colour-science (Python)
- Interfaces
term· Terminalapi· API/SDK- GUI need · task setting
CL· CLI-sufficient- GUI need · ordinary access
CL· CLI-sufficient- Interaction primitives
code· Code & commands- Work product
WP05· Time-based media- Conditions
- None declared
- Proposed difficulty
- 2 / 5
Proposed verifier
Both files parse with the right sizes and bit depth; on 10k random RGB samples, output vs the 17-pt source evaluated tetrahedrally: max ΔE2000 <= 0.5 for both; .3dl input shaper line correct.
Proposed partial credit
30% formats valid; 50% interpolation accuracy; 20% 3dl bit depth and shaper
R06.04.02.06-T17 Chart-based shot matching across six clips
Original instruction
In Resolve, 6 clips from different cameras in input/match_test.drp each start with an X-Rite ColorChecker. Use Color Match (or manual primaries) on the chart frames so every clip matches the reference chart values, then render the 6 chart frames as stills to out/match/.
- Platforms
mac· macOSwin· Windowslin· Linux- Applications
- DaVinci Resolve
- Interfaces
desk· Desktop GUI- GUI need · task setting
G2· GUI-essential- GUI need · ordinary access
G2· GUI-essential- Interaction primitives
click· Point & clickvis· Visual judgmentdrag· Drag & drop- Work product
WP05· Time-based media- Conditions
- None declared
- Proposed difficulty
- 3 / 5
Proposed verifier
Grader samples 24 patches per still: mean ΔE2000 vs reference <= 2.5 and max <= 6 for each clip; neutral patches |a*|,|b*| <= 2. External scripting disabled; fusionscript connections detected -> 0.
Proposed partial credit
60% mean patch error within threshold for all clips; 25% neutrals; 15% max error
R06.04.02.06-T18 Controlled highlight roll-off for a 1000-nit grade
Original instruction
Shots 3-7 of Resolve timeline 'Concert_HDR' (PQ 1000) have stage lights clipping hard. Add a highlight roll-off (soft clip / custom curve in nits) so values above 600 nits compress smoothly into 1000 without banding while midtones (<= 100 nits) change by less than ΔE ITP 2. Render out/concert_fixed.mov.
- Platforms
mac· macOSwin· Windowslin· Linux- Applications
- DaVinci Resolve Studio
- Interfaces
desk· Desktop GUI- GUI need · task setting
G2· GUI-essential- GUI need · ordinary access
G2· GUI-essential- Interaction primitives
click· Point & clickvis· Visual judgmentdrag· Drag & drop- Work product
WP05· Time-based media- Conditions
licensed· Licensed pro app- Proposed difficulty
- 4 / 5
Proposed verifier
Grader: no pixel > 1000 nits; tone curve estimated from source/output pairs is monotonic with continuous slope above 600 nits (no hard knee); midtone ΔE ITP <= 2; banding metric on light gradients below threshold. External scripting disabled; fusionscript connections detected -> 0.
Proposed partial credit
30% no values above peak; 35% smooth roll-off; 20% midtones preserved; 15% no banding
R06.04.02.06-T19 Author an OCIO config with show looks and roles
Original instruction
Write an OCIO v2 config out/show.ocio for the show: ACES 1.3 base (reference ACES2065-1), roles scene_linear=ACEScg, color_timing=ACEScct, compositing_log=ACEScct, two looks 'DayLook' and 'NightLook' from input/looks/*.cube (in ACEScct), and displays sRGB and Rec.1886 with views including the looks. Validate with ociocheck and load it in Nuke.
- Platforms
lin· Linuxmac· macOSwin· Windows- Applications
- OpenColorIO · Nuke
- Interfaces
term· Terminaldesk· Desktop GUI- GUI need · task setting
HY· Hybrid- GUI need · ordinary access
CL· CLI-sufficient- Interaction primitives
code· Code & commandsclick· Point & click- Work product
WP06· Code/engineering- Conditions
licensed· Licensed pro appmultiapp· Multi-app- Proposed difficulty
- 4 / 5
Proposed verifier
ociocheck passes with 0 errors; roles exact; looks process in ACEScct; grader processes test pixels through each display/view and matches reference within 1e-4; Nuke session log shows the config loaded and a Viewer using 'DayLook' view.
Proposed partial credit
30% valid config and roles; 40% looks and views correct numerically; 30% loads and works in Nuke
R06.04.02.06-T20 Three-way deliverables package with QC report
Original instruction
From the graded HDR timeline in input/doc_final.drp deliver: HDR10 PQ ProRes 4444 XQ with metadata, SDR Rec.709 ProRes 422 HQ via a trim pass, and a web H.264 sRGB-tagged file with -16 LUFS audio. Write out/qc.json with each file's color tags, MaxCLL/MaxFALL, loudness and duration.
- Platforms
mac· macOSwin· Windows- Applications
- DaVinci Resolve Studio · ffmpeg
- Interfaces
desk· Desktop GUIterm· Terminal- GUI need · task setting
HY· Hybrid- GUI need · ordinary access
HY· Hybrid- Interaction primitives
click· Point & clickvis· Visual judgmentcode· Code & commands- Work product
WP05· Time-based media- Conditions
licensed· Licensed pro applong· Long-horizonmultiapp· Multi-app- Proposed difficulty
- 5 / 5
Proposed verifier
Grader checks: HDR file tags and SEI; SDR file BT.709 and ΔE2000 <= 3 vs reference SDR frames; web file BT.709/sRGB tag, integrated loudness -16 +/-0.5 LUFS; durations frame-exact; qc.json values within tolerance of grader measurements.
Proposed partial credit
30% HDR master spec; 30% SDR trim accuracy; 20% web file spec and loudness; 20% QC report accurate
Mapped research references
No benchmark reference is assigned to this class in the current mapping.