Natural Colour ProfileAdjustment for Professionals Peter Morovič, Ján Morovič, Jordi Arnabat, Jordi Vilar, Angel Albarrán ...
Understanding Artists• Today’s devices are more repeatable, more consistent and more accurate than ever before … they ar...
Tuning a device (printer)• Calibration: consistent, repeatable output repeatable • All devices produce skin tones of ...
Learn my preference Consistent Accurate 4 /15
Learn my preference Consistent Accurate Mine 5 /15
From image to print ICC profile transformation sequence in a printing pipeline: most photographers use it (and trust it...
Image based workflowelect Image Select Image ...
ICC based workflow Select Image Select Image 1 ...
ICC mechanics• The ICC profile uses a series of sRGB mechanisms to map to/from device sp...
Algorithm• Introduced adjustment shifts Unadjusted Adjusted expressed in PCS (CIE...
Scope• Systematic adjustment • Makes sense if there is a systematic adjustment that is desired by a user – not if eac...
Results – Qualitative adjusted by alg...
Results – Quantitative• Compare pixel–by–pixel CIE DE 2000 5 between: ...
Summary• colour preference algorithm shown to work well for ‘small tweaks’ (on top of calibration and profiling)• uses us...
G. Pinkhassov – MagnumDank u wel!
of 15

Natural Color Profile Adjustment for Professionals

While there may be no point in arguing about taste, creative professionals make a living from sharing theirs. Making specific, individual color preferences that a creative professional knows how to achieve when creating content on a display also propagate into print is a significant challenge since it lacks real-time feedback. The present paper introduces a method for allowing creative professionals to use the tools they know and love to also personalize the color behavior of their devices. This is achieved by analyzing color changes applied to images and applying them to a device’s ICC profile. As a result the personalized device results in customized color behavior regardless of the workflow used. The paper describes the ICC profile transformation algorithm in detail and provides a color error analysis of its performance.
Published on: Mar 3, 2016
Published in: Technology      Business      
Source: www.slideshare.net


Transcripts - Natural Color Profile Adjustment for Professionals

  • 1. Natural Colour ProfileAdjustment for Professionals Peter Morovič, Ján Morovič, Jordi Arnabat, Jordi Vilar, Angel Albarrán Hewlett Packard Company Barcelona, Spain Kindly presented by Hila Nachlieli of HP Labs, Israel The 6th European Conference on Colour in Graphics,Imaging and Vision May 2012, Amsterdam (The Netherlands)
  • 2. Understanding Artists• Today’s devices are more repeatable, more consistent and more accurate than ever before … they are also less personal• Photographers, designers, artists want to have a personal look & feel in all embodiments of their art (screen, print, ...)• Artists care a lot about their art: “Sometimes I have not even recognised my own photographs. I have even hesitated to call them my G. Pinkhassov own. […] Whoever controls the editing of a photographer, controls his fate.” G. Pinkhassov/Magnum 2 /15
  • 3. Tuning a device (printer)• Calibration: consistent, repeatable output repeatable • All devices produce skin tones of the same mine? green hue, neutrals are consistently red• Profiling: accuracy • Skin tones print as skin-tones and neutrals are neutral• What about my preference? accurate [the missing part] mine? • I like my skin-tones to be warmer • I like my neutrals to be bluish • I like slightly higher contrast • ...and I want it systematically 3 /15
  • 4. Learn my preference Consistent Accurate 4 /15
  • 5. Learn my preference Consistent Accurate Mine 5 /15
  • 6. From image to print ICC profile transformation sequence in a printing pipeline: most photographers use it (and trust it) to get best results sRGB preview of image sRGB image in output profile sRGB Printer ICC Profile ICC Profile sRGB LAB dRGB sRGB.icc Printer/Media.icc Q1: how can we modify this workflow?Q2: how can an artist easily “teach” the workflow it’s preference? 6 /15
  • 7. Image based workflowelect Image Select Image 1 Adjust all Adjust only images one one image by one in Photoshop Adjust in Photoshop 2 NO (4) NO (4)iterative iterative justments adjustments3 Satisfied? Print 3 Satisfied? YES (4) YES (4)w ICC Profile Save copy of computed Next Image image incl printd registered adjustments As many A single file files as is created images/ media 7 /15
  • 8. ICC based workflow Select Image Select Image 1 1 Adjust only one image Adjust in Photoshop Adjust in Photoshop 2 2 NO (4) iterative iterative adjustments adjustments Print 3 Satisfied? Print 3 YES (4) YES (4) New ICC Profile Save copy of is computed Next Image image incl print and registered adjustments A single file is created 8 /15
  • 9. ICC mechanics• The ICC profile uses a series of sRGB mechanisms to map to/from device space to/from colorimetry (the PCS), such as: LAB sRGB ICC Profile - 1D linearisation tables - ND Colour Look-Up Tables Desired CLUT - 3x3 transforms (for XYZ PCS) Color Adjustment Reference dRGB ICC Profile• Given the “before” and “after” images adjusted by photographer: CLUT• A pixel-by-pixel and colour-by-colour dRGB New ICC analysis let’s us devise a new ICC profile Profile that embodies the desired adjustment 9 /15
  • 10. Algorithm• Introduced adjustment shifts Unadjusted Adjusted expressed in PCS (CIE LAB) – a sRGB dRGB perceptually near-uniform space AToB AToB & BToA• Uses same interpolation as CMM LAB dRGB• Importance of shift is weighted by number of pixels changed – average in 1D LUT (out 1D LUT)-1 (out 1D LUT)-1 shift encoded LAB dRGBCLUT dRGBCLUT• Can represent some spatial editing• Image-gamut relative colour Weights Shifts adjustments ϖ δ• Post-algorithm LUT smoothing is applied Weighted shifts Δ = Σ(δϖ)/Σ(ϖ)• Fast – image is parsed only once 10 /15
  • 11. Scope• Systematic adjustment • Makes sense if there is a systematic adjustment that is desired by a user – not if each image requires a different change• Does not replace retouching • Spatial changes will be taken into account on average (e.g. if the blues of a sky are modified but the blue of the sea is not, an average will be considered)• Global-local choice • Colour–global adjustments applied to image with native gamut should remain local to image gamut or not?• Designed for fine-tuning • Assumes a calibrated, profiled device on top of which fine adjustments are introduced. Can work to compensate small shortcomings/trade-offs of profiling engines too. 11 /15
  • 12. Results – Qualitative adjusted by algorithmic• Environment: Calibrated monitor (EIZO), unadjusted photographer processing Photoshop CS4, Calibrated and Profiled HP Z3100 (ICC also used for soft-proofing), VeriVide viewing booth with D50 for viewing prints• 24 sets of [photo/substrate] adjustments by 12 pro-photo/fine artists using full Photoshop functionality (USA, Spain, UK) –  after adjustment, new ICC computed and applied: • asked to compare [adjusted image + reference profile] vs [unadjusted image + computed profile] • no difference was reported • asked to apply computed profile to other images from “a set” • adjustments were as expected unadjusted algorithmic processing 12/15
  • 13. Results – Quantitative• Compare pixel–by–pixel CIE DE 2000 5 between: 4.5 [adjusted image + reference profile] vs [unadjusted image + computed profile] 4 3.5 • Median always below 0.5 DE00 – 3 blue circles CIE DE2000 2.5 • 95th %tile always below 2.5 DE00 2 (below 1 DE00 for all but one case) 1.5 – red circles 1• Reference values: 2.5 DE00 noticeable 0.5 – 5 DE00 objectionable in complex 0 1 2 3 4 5 6 7 8 9 10 11 12 1 14 15 16 17 18 19 20 21 22 23 24 photographic content Image Adjustment Number 13 /15
  • 14. Summary• colour preference algorithm shown to work well for ‘small tweaks’ (on top of calibration and profiling)• uses using existing ICC mechanisms (results in a standard ICC profile)• fits seamlessly in existing workflow (known tools to make adjustment, simplified workflow when printing)• demonstrated on printers – applies equally to projectors/displays/... 14 /15
  • 15. G. Pinkhassov – MagnumDank u wel!

Related Documents