Research on how to present images showing uneven foundation application that do not cause unfavorable impressions to the viewer

Research on how to present images showing uneven foundation application that do not cause unfavorable impressions to the viewer Miho Kajita Satoshi Nakamura, Takayuki Itoh Meiji University

Our Goal 2

Our Goal 3

Project Timeline Our past work Binary classification of bare skin and foundation skin images taken with a smartphone using machine learning. Average accuracy: 82.3% Miho Kajita, Satoshi Nakamura. Basic Research on How to Apply Foundation Makeup Evenly on Your Own Face, 20th IFIP TC14 International Conference on Entertainment Computing (IFIP ICEC 2021), pp.402-410, 2021. Goal Proposed system 4

Project Timeline Our past work Binary classification of bare skin and foundation skin images taken with a smartphone using machine learning. Average accuracy: 82.3% Miho Kajita, Satoshi Nakamura. Basic Research on How to Apply Foundation Makeup Evenly on Your Own Face, 20th IFIP TC14 International Conference on Entertainment Computing (IFIP ICEC 2021), pp.402-410, 2021. Goal Proposed system We have not investigated appropriate visualization methods for uneven foundation application. 5

Related Work A system for quantifying the amount of foundation by using an optical filter that emphasizes the difference in the wavelength characteristics of reflection between bare skin and foundation skin [Nishino 2013] 6

Related Work A system for quantifying the amount of foundation by using an optical filter that emphasizes the difference in the wavelength characteristics of reflection between bare skin and foundation skin [Nishino 2013] 7

Related Work Do you want to be visualized like this? :( fying ation filter ence istics skin and foundation skin [Nishino 2013] 8

Purpose To find a comfortable visualization of unevenness by comparing several methods 9

Visualization methods preparation of 4 types of visualizations which visualized foundation unevenness heatmap entirely heatmap partially encircled filled General visualized methods Simple marking methods 10

Visualization methods: color variation 1. entirely red heat map 2. partially red heat map 3. entirely blue heat map 4. partially blue heat map 5. encircled by red 6. encircled by blue 7. encircled by yellow 8. encircled by green 9. filled with red 10. filled with blue 11. filled with yellow 12. filled with green 11

Experiment1: impression To investigate the difference between the impression of visualizing uneven foundation application on one's own face and that on the face of another person. 「静かな」か「うるさい」かを評価してください ○ 非常に静かな ○ とても静かな ○ 少し静かな ○ どちらでもない ○ 少しうるさい ○ とてもうるさい ○ 非常にうるさい 「明るい」か「暗い」かを評価してください ○ 非常に明るい ○ とても明るい ○ 少し明るい ○ どちらでもない ○ 少し暗い ○ とても暗い ○ 非常に暗い 「透明感がある」か「くすんでいる」かを評価してください ○ 非常に透明感がある ○ とても透明感がある ○ 少し透明感がある ○ どちらでもない ○ 少しくすんでいる ○ とてもくすんでいる ○ 非常にくすんでいる 13 adjective pairs e.g. bright ⇔ dark, warm ⇔ cold, active ⇔ passive, radiant ⇔ subdued and more… 12

Result1 Own face : -0.05 vs Other's face : 0.16 13

Result1 Own face : -0.05 vs Other's face : 0.16 the participants had more negative impressions when the unevenness was visualized on their face images than when on the images of other people's faces. 14

Result1 Something like thermography is visualized on the face of the participant and not the participant themselves but researchers evaluate the visualization result. conventional visualization studies It is necessary to consider the impression that the person receives from the visualization of uneven foundation on their own face. Proposed system 15

Experiment2: Favorability and recognizability Comparison of 36 methods in a round-robin fashion. e.g. Which do you like better? Which is easier to check unevenness? 「左の方が好ましい」か「右の方が好ましい」かを評価してください ○ 左の方が好ましい ○ 右の方が好ましい 「左の方が塗りムラがわかりやすい」か「右の方が塗りムラがわかりやすい」かを評価してください ○ 左の方が塗りムラがわかりやすい ○ 右の方が塗りムラがわかりやすい 評価する 16

Result2 Dislike like 17

Discussion Good method • makes it easy to see the relative position of the unevenness of the paint • with a gorgeous impression Bad method • Resemble to disease symptoms, such as rashes and bruises. 18

Prototype System 19

Prototype System 👍Easy to understand because of real-time confirmation 👎 Time lag. Accuracy is not good 👍The display format and color of the uneven paint was easy to see 20

summary To compare visualization methods for uneven makeup application purpose circled by and filled with green methods is highly rated result 01 background It is difficult to check unevenness of foundation application 02 03 experiences Investigation of impression and the most preferred method 04 Thank you for attention! 21