Analysis of Average Hand-drawing and Its Application

Analysis of Average Hand-drawing and Its Application Shinjiro Niino (Meiji University) Nanae Hagiwara, Satoshi Nakamura, Masaaki Suzuki, and Takanori Komatsu

Average Painter 1/36 Background > Purpose > Method > Experiment > Application > Conclusion ※ Please come on my booth in demonstration session

Background

Background 2/36 Background > Purpose > Method > Experiment > Application > Conclusion • When was the last time you drew illustrations? • 88% of people cannot be satisfied with their hand-drawing Do you have a feeling of being not good at handwritings or hand-drawings (ZEBRA - http://www.zebra.co.jp/press/news/2014/0918.html)

Background 2/36 Background > Purpose > Method > Experiment > Application > Conclusion • When was the last time you drew illustrations? • 88%People of people cannot be satisfied have desires for with their hand-drawing better hand-drawings Do you have a feeling of being not good at handwritings or hand-drawings (ZEBRA - http://www.zebra.co.jp/press/news/2014/0918.html)

Related Work 3/36 Background > Purpose > Method > Experiment > Application > Conclusion • ShadowDraw [Lee 2011] − guides userʼs drawings by displaying a shadow image of target illustration in background

Related Work 3/36 Background > Purpose > Method > Experiment > Application > Conclusion • ShadowDraw [Lee 2011] − guides userʼs drawings by displaying a shadow image of target illustration in background User cannot draw the target illustration if it is not in the database

Related Work 4/36 Background > Purpose > Method > Experiment > Application > Conclusion • Pegasus [Igarashi 1994] − support to draw geometric design easily and rapidly by suggesting correct strokes

Related Work 4/36 Background > Purpose > Method > Experiment > Application > Conclusion • Pegasus [Igarashi 1994] − support to draw geometric design easily and rapidly by suggesting correct strokes This method may eliminate users’ unique personality of hand-drawings

Background 5/36 Background > Purpose > Method > Experiment > Application > Conclusion • There is not enough knowledge about the characteristics of handdrawing in previous work

Purpose of Study 6/36 Background > Purpose > Method > Experiment > Application > Conclusion • To clarify characteristics of handdrawn illustrations • To realize the supporting system for satisfactory hand-drawings based on the characteristics

Study about Handwritings 7/36 Background > Purpose > Method > Experiment > Application > Conclusion • Average handwritten Hiragana-characters are beautiful [Nakamura 2016] Hiragana-characters

Study about Handwritings 8/36 Background > Purpose > Method > Experiment > Application > Conclusion • The average handwritten characters were highly evaluated than daily ones Comparison of Daily handwritten character with Average character 1day , , 2day ,・・・・, 3day < 4day 5day + Average Character + + ・・・・ + n

Study about Handwritings 9/36 Background > Purpose > Method > Experiment > Application > Conclusion • The average handwritten characters of all users were highly evaluated than the ones of each user Comparison of Each user’s average character with All users’ average character , , ,・・・・ , < + + + ・・・・+ n

Questions about Hand-drawings 10/36 Background > Purpose > Method > Experiment > Application > Conclusion • Q1: Are the average illustrations beautiful? • Q2: Are the average illustrations of several users beautiful? • Q3: Are illustrations by dominant hand similar to ones by nondominant hand? Or Not?

Hypotheses 11/36 Background > Purpose > Method > Experiment > Application > Conclusion • Hypo1: The average illustrations are more beautiful than original ones , ,・・・・, , < + + + ・・・・ + n • Hypo2: The average illustrations of all users are more beautiful than the ones of each user , , , ・・・・ , + + + ・・・・ + , < + + + ・・・・ + n • Hypo3: The average illustrations by non-dominant hand are similar to the ones by dominant hand n ≒ + + + ・・・・ + n

Method

Averaging Method 12/36 Background > Purpose > Method > Experiment > Application > Conclusion • Mathematize strokes of illustrations by using Fourier Series Expansion 𝑥= 272.08cos(0t) + 0sin(0t) + 16.62cos(1t) + 0.09sin(1t) + 29.81cos(2t) + -0.32sin(2t) + 2.91cos(3t) + 0.05sin(3t) + -37.67cos(4t) + 0.8sin(4t) + 18.93cos(5t) + -0.5sin(5t) + ・・・ 𝑦= 229.95cos(0t) + 0sin(0t) + 42.82cos(1t) + 0.23sin(1t) + 4.75cos(2t) + -0.05sin(2t) + 44.24cos(3t) + -0.7sin(3t) + 16.89cos(4t) + 0.36sin(4t) + -2.51cos(5t) + 0.07sin(5t) + ・・・

Averaging Method 13/36 Background > Purpose > Method > Experiment > Application > Conclusion • Average the mathematized strokes 𝑖=2 𝑖=1 𝑖=3 𝑵 𝟏 𝒙𝒊 = . 𝒇𝒊,𝒌 𝒕 𝑵 𝒌3𝟏 𝑵 𝑁 ・ ・ ・ 𝟏 𝒚𝒊 = . 𝒈𝒊,𝒌 𝒕 𝑵 𝒌3𝟏

Averaging Method 13/36 Background > Purpose > Method > Experiment > Application > Conclusion • Average the mathematized strokes 𝑖=2 𝑖=1 𝑖=3 𝑵 𝟏 𝒙𝒊 = . 𝒇𝒊,𝒌 𝒕 𝑵 𝒌3𝟏 𝑵 𝑁 ・ ・ ・ 𝟏 𝒚𝒊 = . 𝒈𝒊,𝒌 𝒕 𝑵 𝒌3𝟏

Demonstration of Method 14/36 Background > Purpose > Method > Experiment > Application > Conclusion

Experiments

Experiments 15/36 Background > Purpose > Method > Experiment > Application > Conclusion • We conducted experiments − Created a dataset of hand-drawn illustrations − Evaluated hand-drawn illustrations Experiment1 ・・・・ Experiment2 Experiment3 ・・・・ + + n + ・・・・ + < + < + ≒ + + n + + ・・・・ + + ・・・・ + n + + ・・・・ + n

Dataset 16/36 Background > Purpose > Method > Experiment > Application > Conclusion Participants Illustrations Times 8 people Student / Age : 20-22 / Men:4 Women:4 4 illustrations Dominant hand 5times Non-dominant hand 5times

Dataset 17/36 Background > Purpose > Method > Experiment > Application > Conclusion • We generated average illustrations 1 2 3 A E B F C G D H 4 5 Average User’s illustrations Average Each user’s average illustration All users’ average illustration

Experiments 18/36 Background > Purpose > Method > Experiment > Application > Conclusion Experiment1 ・・・・ Experiment2 Experiment3 ・・・・ + + n + ・・・・ + < + < + ≒ + + n + + ・・・・ + + ・・・・ + n + + ・・・・ + n

Experiments 19/36 Background > Purpose > Method > Experiment > Application > Conclusion Experiment1 ・・・・ Experiment2 Experiment3 ・・・・ + + n + ・・・・ + < + < + ≒ + + n + + ・・・・ + + ・・・・ + n + + ・・・・ + n

Ex1: ・・・・ < + + + ・・・・ + n 20/36 Background > Purpose > Method > Experiment > Application > Conclusion • Rank illustrations from 1st to 3rd − Evaluators : 16 people

Result1: ・・・・ < + + 21/36 + ・・・・ + n Background > Purpose > Method > Experiment > Application > Conclusion + +...+ n Average 𝒏 𝐭imes illustration 𝒏 times illustration by dominant hand by non-dominant hand illustrations

Result1: ・・・・ < + + 21/36 + ・・・・ + n Background > Purpose > Method > Experiment > Application > Conclusion + +...+ n Average 𝒏 𝐭imes illustration 𝒏 times illustration by dominant hand by non-dominant hand illustrations Average illustrations are more beautiful than original ones

Average Illustrations 22/36 Background > Purpose > Method > Experiment > Application > Conclusion • The average of several hand-drawn illustrations Original Average 1 3 2 4 5 Average

Experiments 23/36 Background > Purpose > Method > Experiment > Application > Conclusion Experiment1 ・・・・ Experiment2 Experiment3 ・・・・ + + n + ・・・・ + < + < + ≒ + + n + + ・・・・ + + ・・・・ + n + + ・・・・ + n

Ex2: ・・・・ < + + + ・・・・+ n 24/36 Background > Purpose > Method > Experiment > Application > Conclusion • Rank illustrations from 1st to 3rd − Evaluators : 16 people

Result2: ・・・・ < + + + ・・・・ + n 25/36 Background > Purpose > Method > Experiment > Application > Conclusion The average illustration of all users The average illustrations of each user

Result2: ・・・・ < + + + ・・・・ + n 25/36 Background > Purpose > Method > Experiment > Application > Conclusion The average illustration of all users The average illustrations of each The average illustrations of all useruser are more beautiful than the ones of each user

Average Illustration 26/36 Background > Purpose > Method > Experiment > Application > Conclusion • The average illustrations of several users A B C (A+B+C)/3 Average

Experiments 27/36 Background > Purpose > Method > Experiment > Application > Conclusion Experiment1 ・・・・ Experiment2 Experiment3 ・・・・ + + n + ・・・・ + < + < + ≒ + + n + + ・・・・ + + ・・・・ + n + + ・・・・ + n

Ex3: + + + ・・・・ + n ≒ + + + ・・・・ + n 28/36 Background > Purpose > Method > Experiment > Application > Conclusion • Make 8 pairs that they thought similar illustration from average illustrations by dominant hand and non-dominant hand

Result3: + + + n ・・・・ + ≒ + + + n ・・・・ + 29/36 Background > Purpose > Method > Experiment > Application > Conclusion

Result3: + + + n ・・・・ + ≒ + + + n ・・・・ + 29/36 Background > Purpose > Method > Experiment > Application > Conclusion The average illustrations by nondominant are similar to the ones by dominant hand

Result3: + + + ・・・・ + n ≒ + + + n ・・・・ + 30/36 Background > Purpose > Method > Experiment > Application > Conclusion The average illustrations by dominant hand The average illustrations by non-dominant hand A A B B C D ≒ C D

Clarified Characteristics 31/36 Background > Purpose > Method > Experiment > Application > Conclusion 1. The average illustrations are more beautiful than the original ones 2. The average illustrations of all users are more beautiful than the ones of each users 3. The average illustrations by non-dominant hand are similar to the ones by dominant hand

Average Painter 32/36 Background > Purpose > Method > Experiment > Application > Conclusion • Prototype of the system to support hand-drawings by averaging strokes l Redrawn Stroke l Undone Stroke

Demonstration Video 33/36 Background > Purpose > Method > Experiment > Application > Conclusion ※ Please come on my booth in demonstration session

Conclusion

Conclusion(1) 34/36 Background > Purpose > Method > Experiment > Application > Conclusion • We made 3 hypotheses about hand-drawn illustrations, and clarified characteristics of ones Hypo1 ・・・・ Hypo2 Hypo3 ・・・・ + + n + ・・・・ + < + < + ≒ + + n + + ・・・・ + + ・・・・ + n + + ・・・・ + n

Conclusion(2) 35/36 Background > Purpose > Method > Experiment > Application > Conclusion • We implemented a prototype of drawing system to support drawing by averaging hand-drawings Average Functions

Future Work 36/36 Background > Purpose > Method > Experiment > Application > Conclusion • To conduct evaluate experiments about Average Painter • To apply the result of hypothesis2,3 − Hypothesis2: n Collaborate in drawing with usersʼ friends or children or favorite comic artists... − Hypothesis3: n n We clarified a gap of usersʼ ideal illustrations and target illustrations Realize a system to improve usersʼ ideal illustrations