5. Definition of Font Elements
The definition of TR X 003:2000 includes 14 elements for capital roman letters. We added 5
elements and distinguished direction or position of elements, finally defined 31 elements.
Kaori Yoshida <kaori@ai.kyutech.ac.jp>
7. Definition of Font Element Parameters
Kaori Yoshida <kaori@ai.kyutech.ac.jp>
8. Example of Font Element Parameters
For example, the element stem (<Stem>) of upper right point (position=URP) is started at the
point (195, 131) and connected to the point (324, 462) with vertical line (vline).
Kaori Yoshida <kaori@ai.kyutech.ac.jp>
9. Definition of Font Element Attributes
The changes of an element affect to all letters having the same element for total design.
Kaori Yoshida <kaori@ai.kyutech.ac.jp>
10. Examples of Font Element attributes
Font Element Attribute Description
R_Horn
enlarge_R_Horn make bigger R_Horn
R_Horn shrink_R_Horn make smaller R_HornR_Horn
mutate_R_Horn change shape R_Horn
Spur
sharpen_Spur make sharper Spur
Spur enlarge_Spur make bigger SpurSpur
mutate_Spur change shape Spur
Stress
fluff_Stress make fluffy Stress
Stress jag_Stress make jagged StressStress
mutate_Stress change shape Stress
Each font element has each different font element attributes.
We defined 101 font element attributes totally.
Kaori Yoshida <kaori@ai.kyutech.ac.jp>
13. Font Creation Application PALETTE
PALETTE (Personal Adapted LETTEr)
•create new font candidates base on general font, or
stored font characteristics with Kansei label
•generate new fonts using genetic algorithm which can
modify each element of the selected candidates
•export font image and save font characteristics
Kaori Yoshida <kaori@ai.kyutech.ac.jp>
14. Why based on IGA?
IGA can ...
•reflect user's Kansei interactively
•change of font element shape can reflect to same element
of the other letters
•the experimental results include a possibility to analyze
relation between users Kansei and font elements based
on the users activities history
Kaori Yoshida <kaori@ai.kyutech.ac.jp>
15. IGA
•Chromosomes are designed by history of operated font
element attributes. The history is represented by font
elements, font element parameters, and font element
attributes of each mutation.
•We encoded font elements and designed users feedback,
shown in the selection of candidates by users, as fitness
function for genetic algorithm.
•The first 8 candidates are created by mutation of font
elements selected randomly.
Kaori Yoshida <kaori@ai.kyutech.ac.jp>
24. Task
(1) perform evolution with 1 or 2 candidates
(2) create the font suit to your impression
(3) save label
(4) save font as image
Kaori Yoshida <kaori@ai.kyutech.ac.jp>
25. Questions
(A) PALETTE provides the various fonts?
(B) PALETTE provides the fonts suit to your Kansei?
(C) PALETTE provides the practical fonts?
(D) PALETTE is easy to use?
(E) PALETTE creates the fonts with your satisfaction?
(F) PALETTE gives you heavy load to create fonts?
Kaori Yoshida <kaori@ai.kyutech.ac.jp>
26. Perspective of Question Answer # Comments
(A) Variety
Yes 9 / 10
-
The parameter mutation intensity reflects variety
directly.
(A) Variety
No 1 / 10
-
Itʼs hard to observe the variety of candidates if the
parameter mutation intensity and frequency set
with low value.
-
There are very strong limitations to create various
fonts.
(B) Kansei reflection
Yes 7 / 10
-
The candidates closed to my favorite shape step
by step.
(B) Kansei reflection
No 3 / 10
-
The candidates never included rounded shape,
which gives soft impression.
-
Itʼs hard to reflect each element changes to total
design of each letter.
(C) Practical to use
Yes 8 / 10
-
Itʼs good to create peculiar fonts.
-
Itʼs useful to make poster.(C) Practical to use
No 2 / 10 -
I cannot imagine purpose to use.
Answers of the questionnaire (A) (B) (C)
Kaori Yoshida <kaori@ai.kyutech.ac.jp>
27. Perspective of Question Answer # Comments
(D) Easy to use
Yes 6 / 10 -
It has good interface to use.
(D) Easy to use
No 4 / 10
-
Itʼs inconvenient we cannot back to the previous
page.
-
Itʼs hard to find which part of letter changed.
-
The application was getting slower to process.
(E) Satisfaction
Yes 5 / 10
-
I could create the fonts like calligraphy, like bone,
with the impression elegant and soft.(E) Satisfaction
No 5 / 10 -
I could not create rounded or light shape font.
(F) Heavy load
Yes 6 / 10
-
It took long time and many steps (generations) to
get my favorite font.(F) Heavy load
No 4 / 10 -
I could use the application without any heavy load.
Answers of the questionnaire (D) (E) (F)
Kaori Yoshida <kaori@ai.kyutech.ac.jp>