webinar: reproducible display sparkle measurements · webinar: reproducible display sparkle...

Webinar: Reproducible Display

Sparkle Measurements

TechnoTeam Bildverarbeitung GmbH,

Werner-von-Siemens-Str. 5, 98693, Ilmenau, Germany

Elektrobit Automotive GmbH,

Lise-Meitner-Straße 10, 89081, Ulm, Germany

Volkswagen AG,

38436, Wolfsburg, Germany

Ingo Rotscholl, Development Engineer, PhD

• PhD in 2017 in the field of photometry at the Light Technology

Institute (LTI) of Karlsruhe Institute of Technology (KIT)

• Development engineer for display metrology at TechnoTeam

Bildverarbeitung GmbH since 2018

• Several publications on display metrology including the topics

uniformity, sticking images, AR/VR and automotive displays

• Member in national and international standardization

organizations such as IEC, ICDM, DKE and VDI

About the speaker

Table of contents

• Measuring sparkle fundamentals

– Introduction

– Separation of sparkle

– Reproducibility

• Measurement setup

– Setup problem 1: Initial focus position

– Setup problem 2: Dust, defects ...

– Setup problem 3: Reproduction scale

– Setup problem 4: Measurement field angle

• Conclusions, summary and Q/A

Table of contents

• Measuring sparkle fundamentals

– Introduction

– Separation of sparkle

– Reproducibility

• Measurement setup

– Setup problem 1: Initial focus position

– Setup problem 2: Dust, defects ...

– Setup problem 3: Reproduction scale

– Setup problem 4: Measurement field angle

• Conclusions, summary and Q/A

Introduction: Sparkle and high quality displays

Self emissive display

▪ Anti Glare Layer (AGL) enhance contrast and readability in direct

lighting environments (e.g. sun)

▪ Anti Glare Layer may blur image (pixel cross talk)

▪ Anti Glare Layer may lead to high frequency non-uniformity (sparkle)

▪ Not yet standardized (in progress)

▪ Reproducibility is a main issue

Introduction: Sparkle and high quality displays

▪ BlackMURA compliant setup

▪ Measurement without removing AGL

Sparkle measurements: TechnoTeam conditions:

Separation of sparkle

• Influence of periodic pixel structure

Bare pixel matrix Pixel matrix with sparkle Sparkle without pixel matrix

Separation of sparkle

Known methods [2]

• De-focus [3,4]

• Filter in spatial

domain [1,5]

• Filter in frequency

domain [2,6]

• Undersampling [7,8]

Pixel matrix with sparkle

• Separation of periodic pixel structure from sparkle in luminance picture

Sparkle without pixel matrix

Separation of sparkle

Known methods [2]

• De-focus [3,4]

• Filter in spatial

domain [1,5]

• Filter in frequency

domain [2,6]

• Undersampling [7,8]

Pixel matrix with sparkle

• Separation of periodic pixel structure from sparkle in luminance picture

Sparkle without pixel matrixReproducibility

issues

Separation of sparkle

Known methods [2]

• De-focus [3,4]

• Filter in spatial

domain [1,5]

• Filter in frequency

domain [2,6]

• Undersampling [7,8]

Pixel matrix with sparkle

• Separation of periodic pixel structure from sparkle in luminance picture

Sparkle without pixel matrixBlackMURA

setup issue(depends)

Separation of sparkle

Known methods [2]

• De-focus [3,4]

• Filter in spatial

domain [1,5]

• Filter in frequency

domain [2,6]

• Undersampling [7,8]

Pixel matrix with sparkle

• Separation of periodic pixel structure from sparkle in luminance picture

Sparkle without pixel matrixOK

Filter in frequency domain

Spatial domain

Spatial domain Amplitudes (FFT)

Filter in frequency domain

Filtered imageFrequency filter

Spatial domain Amplitudes (FFT)

Filter in frequency domain

𝑆𝑝𝑎𝑟𝑘𝑙𝑒 =𝑠𝑡𝑑 𝐿

𝑚𝑒𝑎𝑛 𝐿

Verification of method

• Verification in one reference setting

• Replacement of AG layers with different perceived sparkle

noglass

L1 L2 M1 M2 M3 M4 M5 M6 H1 H2 H3 H4

0%

2%

4%

6%

8%

10%

12%

14%

16%

Sp

ark

le

Expert rating

Display 1 (224ppi) and 12 AGL + (no glass)

Measurement procedure

Alignment according to BlackMURA

Original luminance image

Focus setting in pixel plane (individual pixels)

Filtered luminance image

𝑆𝑝𝑎𝑟𝑘𝑙𝑒 =𝑠𝑡𝑑 𝐿

𝑚𝑒𝑎𝑛 𝐿

Camera Lens Glass

1MP 25mm no glass, L1, L2, M2, M6, H1, H3

50mm no glass, L1, L2, M2, M6, H1, H3

5MP 25mm no glass, L1, L2, M6, H1, H3

50mm no glass, L1, L2, M2, M6, H1, H3

12MP 16mm no glass, L1, L2, M2, M6, H1, H3

25mm no glass, L1, L2, M2, M6, H1, H3

50mm no glass, L1, L2, M1, M2, M3, M4, M5, M6,

H1, H2, H3, H4

5MP* 25mm no glass, L1, L2, M1, M3, M4, M5, H3, H4

Reproducibility

* Different camera

Reproducibility

noglass

L1 L2 M2 M6 H1 H3

0%

2%

4%

6%

8%

10%

12%

14%

Sp

ark

le

16mm 12 MP

25mm 1MP

25mm 5MP

25mm 12MP

50mm 1MP

50mm 5MP

50mm 12MP

Reproducibility

Glass/

Method

No

Glass

L1 L2 M2 M6 H1 H3

CV, direct

measurement25,3% 8,7% 5,8% 8,4% 8,7% 9,7% 8,4%

CV =𝑠𝑡𝑑 𝑆𝑝𝑎𝑟𝑘𝑙𝑒

𝑚𝑒𝑎𝑛 (𝑆𝑝𝑎𝑟𝑘𝑙𝑒)

Measure for relative error:

Measurement setupSetup problem 1: Initial focus position

Focus reproducibility / Depth of focus

AG layer

Pixel layer

Focus reproducibility / Depth of focus

Glass M6

Focus reproducibility / Depth of focus

Glass M6

Sparkle = Maximum

(Optimized focus)

Sparkle = Sparkle(0)

(Initial focus)

Focus reproducibility / Depth of focus

Pix

el la

yer

AG

layer

Depth of focus Circle of confusion is smaller

than sensor pixel size

Size of

sensor pixel

Focus reproducibility / Depth of focus

Pix

el la

yer

AG

layer

Depth of focus Circle of confusion is smaller

than sensor pixel size

• Limitation of focus reproducibility

• Where is the “sparkle plane” Size of

sensor pixel

Focus reproducibility / Depth of focus

Glass M6 Pix

el la

ye

r

AG

laye

r

Depth of focus

Depth of focus

„Sparkle“ plane

Precision/Uncertainty

of manual/auto focus

noglass

L1 L2 M2 M6 H1 H3

0%

2%

4%

6%

8%

10%

12%

14%

Sp

ark

le

Optimized focus

16mm 12MP

25mm 5MP*

25mm 1MP

25mm 5MP

25mm 12MP

50mm 1MP

50mm 5MP

50mm 12MP

Reproducibility with distance focus scan

Reproducibility experiments overview

Glass/

Method

No

Glass

L1 L2 M2 M6 H1 H3

CV, direct

measurement25,3% 8,7% 5,8% 8,4% 8,7% 9,7% 8,4%

CV, optimized

focus18,9% 10,1% 8,6% 3,3% 4,4% 5,3% 4,3%

CV =𝑠𝑡𝑑 𝑆𝑝𝑎𝑟𝑘𝑙𝑒

𝑚𝑒𝑎𝑛 (𝑆𝑝𝑎𝑟𝑘𝑙𝑒)

Measurement setupSetup problem 2: Handling issues

„local“ Camerapixels• Dust, finger prints

• Defects, markers 𝑆𝑝𝑎𝑟𝑘𝑙𝑒𝑙𝑜𝑐𝑎𝑙 =𝑠𝑡𝑑 𝐿𝑙𝑜𝑐𝑎𝑙𝑚𝑒𝑎𝑛 𝐿𝑙𝑜𝑐𝑎𝑙

Handling issues

Filtered image

Handling issues

Imperfections affect standard deviation

• Dust, finger prints

• Defects, markers

Filtered image Local standard deviation matrix

→“Sparkle” matrix

𝑆𝑝𝑎𝑟𝑘𝑙𝑒 = 𝑚𝑒𝑑𝑖𝑎𝑛(𝑆𝑝𝑎𝑟𝑘𝑙𝑒𝑀𝑎𝑡𝑟𝑖𝑥)

Median of sparkle matrix statistically robust

against outliers caused by tiny imperfections

noglass

L1 L2 M2 M6 H1 H3

0%

2%

4%

6%

8%

10%

12%

14%

Sp

ark

le

Optimized focus, local evaluation

16mm_12MP

25mm 5MP*

25mm_1MP

25mm,5MP

25mm,12MP

50mm_1MP

50mm,5MP

50mm,12MP

Reproducibility with distance focus scan and local

evaluation

Reproducibility experiments summary

Glass/

Method

No

Glass

L1 L2 M2 M6 H1 H3

CV, Direct

Measurement25,3% 8,7% 5,8% 8,4% 8,7% 9,7% 8,4%

CV, Optimized

focus18,9% 10,1% 8,6% 3,3% 4,4% 5,3% 4,3%

CV, Local,

optimized focus4,1% 4,1% 4,0% 4,6% 2,8% 5,2% 4,2%

CV =𝑠𝑡𝑑 𝑆𝑝𝑎𝑟𝑘𝑙𝑒

𝑀𝑊 (𝑆𝑝𝑎𝑟𝑘𝑙𝑒)Optimization of reproducibility

Reproducibility experiments summary

Glass/

Method

No

Glass

L1 L2 M2 M6 H1 H3

CV, Direct

Measurement25,3% 8,7% 5,8% 8,4% 8,7% 9,7% 8,4%

CV, Optimized

focus18,9% 10,1% 8,6% 3,3% 4,4% 5,3% 4,3%

CV, Local,

optimized focus4,1% 4,1% 4,0% 4,6% 2,8% 5,2% 4,2%

CV =𝑠𝑡𝑑 𝑆𝑝𝑎𝑟𝑘𝑙𝑒

𝑀𝑊 (𝑆𝑝𝑎𝑟𝑘𝑙𝑒)Optimization of reproducibility

Measurement setupSetup problem 3: Reproduction Scale

• Higher RPS → higher sparkle value and higher focus sensitivity

• Depth of focus depends on lens/measurement distance

Reproduction scale

• Higher RPS → Lower measurement distance with same lens

• Lower measurement distance → Lower depth of focus

Reproduction scale

Reproduction scale

Glass/RPS No

Glass

L1 L2 M2 M6 H1 H3

RPS=3.6 10,25% 6,68% 6,89% 11,13% 6,00% 2,60% 4,61%

RPS=3.0 8,00% 3,80% 3,54% 6,63% 3,24% 2,64% 2,13%

RPS=2.45 4,05% 4,13% 4,04% 4,63% 2,77% 5,17% 4,24%

RPS=2.1 5,66% 4,11% 5,63% 2,38% 4,29% 6,05% 5,24%

• Sensitivity of RPS→ lower RPS more stable

• Sensitivity of RPS→ lower RPS more stable

Reproduction scale

Glass/RPS No

Glass

L1 L2 M2 M6 H1 H3

RPS=3.6 10,25% 6,68% 6,89% 11,13% 6,00% 2,60% 4,61%

RPS=3.0 8,00% 3,80% 3,54% 6,63% 3,24% 2,64% 2,13%

RPS=2.45 4,05% 4,13% 4,04% 4,63% 2,77% 5,17% 4,24%

RPS=2.1 5,66% 4,11% 5,63% 2,38% 4,29% 6,05% 5,24%

Measurement setupSetup problem 4: Measurement field angle

Measurement field angle

𝜗D

d

𝜗𝐻

𝜗𝑉

→“Sparkle” matrix

Measurement field angle

→“Sparkle” matrix

Sparkle increases with increasing field angle

𝜗𝐻

𝜗𝑉

0,0%

2,0%

4,0%

6,0%

8,0%

10,0%

no glass L1 L2 M2 M6 H1 H3

16mm 12MP RPS=2.1

ϑMax=26° ϑMax=14°

ϑMax=7° ϑMax=3°

0,0%

2,0%

4,0%

6,0%

8,0%

10,0%

no glass L1 L2 M2 M6 H1 H3

16mm 12MP RPS=2.1

ϑMax=26° ϑMax=14°

ϑMax=7° ϑMax=3°

Measurement field angle

→“Sparkle” matrix

Sparkle increases with increasing field angle

𝜗𝐻

𝜗𝑉

Measurement setup

• Linear stage

• Vertical/Horizontal alignment

• Low cost solution

• LabSoft Integration

LMK Motor AddOn

• High speed and

high precision alignment

• Enables measuring a large

data base as foundation

for a specification

• Enables sensitivity studies

Measurement setup

Setup Camera Lens RPS1 (Distance) Glass

1-4 1MP 25mm 2,1; 2,45; 3,0; 3,6 no glass, L1, L2, M2, M6, H1, H3

5-8 50mm 2,1; 2,45; 3,0; 3,6 no glass, L1, L2, M2, M6, H1, H3

9-12 5MP 25mm 2,1; 2,45; 3,0; 3,6 no glass, L1, L2, M6, H1, H3

13-16 50mm 2,1; 2,45; 3,0; 3,6 no glass, L1, L2, M2, M6, H1, H3

17-20 12MP 16mm 2,1; 2,45; 3,0; 3,6 no glass, L1, L2, M2, M6, H1, H3

21-24 25mm 2,1; 2,45; 3,0; 3,6 no glass, L1, L2, M2, M6, H1, H3

25-28 50mm 2,1; 2,45; 3,0; 3,6 no glass, L1, L2, M1, M2, M3, M4, M5,

M6, H1, H2, H3, H4

29-32 5MP* 25mm 2,1; 2,45; 3,0; 3,6 no glass, L1, L2, M1, M3, M4, M5, H3,

H4

* Different camera

Data base for specification

Setup Camera Lens RPS1 (Distance) Glass

1-4 1MP 25mm 2,1; 2,45; 3,0; 3,6 no glass, L1, L2, M2, M6, H1, H3

5-8 50mm 2,1; 2,45; 3,0; 3,6 no glass, L1, L2, M2, M6, H1, H3

9-12 5MP 25mm 2,1; 2,45; 3,0; 3,6 no glass, L1, L2, M6, H1, H3

13-16 50mm 2,1; 2,45; 3,0; 3,6 no glass, L1, L2, M2, M6, H1, H3

17-20 12MP 16mm 2,1; 2,45; 3,0; 3,6 no glass, L1, L2, M2, M6, H1, H3

21-24 25mm 2,1; 2,45; 3,0; 3,6 no glass, L1, L2, M2, M6, H1, H3

25-28 50mm 2,1; 2,45; 3,0; 3,6 no glass, L1, L2, M1, M2, M3, M4, M5,

M6, H1, H2, H3, H4

29-32 5MP* 25mm 2,1; 2,45; 3,0; 3,6 no glass, L1, L2, M1, M3, M4, M5, H3,

H4

* Different camera

Data base for specification

Reproducibility experiment II

NoGlass

L1 L2 M4 M1 M3 M5 H3 H4

0%

2%

4%

6%

8%

10%

12%

14%

16%

18%

Sp

ark

le

25mm 1MP

50mm 1MP

25mm 5 MP

50mm 5MP

16mm 12MP

25mm 12MP

50mm 12MP

Display 2 (183 ppi)

CV 2,8% 4,8% 2,4% 2,8% 2,6% 2,7% 1,7% 2,7% 1,8%

Conclusions and summary

Conclusions and summary

• Development of reproducible sparkle measurement specification

(Based on more then 10.000 individual luminance images)

• Reproducibility optimization:

→ Focus/Sparkle scan → To avoid blurring of high/medium sparkle valuesA

G la

yer

Pix

ella

yer

Conclusions and summary

• Development of reproducible sparkle measurement specification

(Based on more then 10.000 individual luminance images)

• Reproducibility optimization:

→ Focus/Sparkle scan → To avoid blurring of high/medium sparkle values

→ Local evaluation → lTo avoid impact of tiny imperfections on low sparkle values

Sparkle matrix

Median

Conclusions and summary

• Development of reproducible sparkle measurement specification

(Based on more then 10.000 individual luminance images)

• Reproducibility optimization:

→ Focus/Sparkle scan → To avoid blurring of high/medium sparkle values

→ Local evaluation → lTo avoid impact of tiny imperfections on low sparkle values

→ Distance/Lens/Camera flexibility enables BlackMURA compliant setup

𝑆𝑝𝑎𝑟𝑘𝑙𝑒 = 8% ± 4% ≈ ± $900,000

Factor Reproduction Scale Field Angle

Influence ++ +

Specification Fixed Value Boundary

• Development of reproducible sparkle measurement specification

(Based on more then 10.000 individual luminance images)

• Reproducibility optimization:

→ Focus/Sparkle scan → To avoid blurring of high/medium sparkle values

→ Local evaluation → lTo avoid impact of tiny imperfections on low sparkle values

→ Distance/Lens/Camera flexibility enables BlackMURA compliant setupPlease contact

us for more

information

Conclusions and summary

𝑆𝑝𝑎𝑟𝑘𝑙𝑒 = 8% ± 1% ≈ ± $50,000

Factor Reproduction Scale Field Angle

Influence ++ +

Specification Fixed Value Boundary

List of references

[1] M. E. Becker, J. Soc. Inf. Disp. 23, p.474 (2015)

[2] M. E. Becker, SID Symposium Digest of Technical Papers. Vol. 49. No. 1. (2018)

[3] M. Scholz, electronic displays Conference (2018)

[4] V. F. Paz, SID Symposium Digest of Technical Papers. Vol. 49. No. 1. (2018)

[5] J. Gollier, SID Symposium Digest of Technical Papers. Vol. 44. No. 1 (2013)

[6] T.W. Hsu, International Display Workshop. (2014)

[7] M. Hayashi, International Display Workshop. (2017)

[8] M. Kurashige, International Display Workshop. (2019)

[9] M. Isshiki, SID Symposium Digest of Technical Papers. Vol. 50. No. 1. (2019)

Thank you for your attention!

Contact: Dr.-Ing. Udo Krüger,

Dr.-Ing. Ingo Rotscholl

Julia Brinkmann

TechnoTeam Bildverarbeitung GmbH

Dr. Christoph Rickers

Volkswagen AG

Jens Rasmussen

Elektrobit Automotive GmbH

Thank you for your attention!

Contact: Dr.-Ing. Udo Krüger,

Dr.-Ing. Ingo Rotscholl

TechnoTeam Bildverarbeitung GmbH

W.-v.-Siemens-Str. 5

D-98693 Ilmenau

www.technoteam.de

Udo.Krueger@technoteam.de

Ingo.rotscholl@technotean.de