401
edits
Changes
under construction
'''Typographics''', in [[Visual Experiments Lain]], is defined as a compound word comprised of the words ''type'' and ''graphic''. It is used to describe scenes in which words float across the screen, usually in accompaniment to [[crosstalk]] or scenes in which [[Lain]] is diving in the [[Wired]]. Such scenes were created by [[Ueda Yasuyuki]] using Adobe After Effects and other programs.
For the purposes of this article, typographics includes any situation where text is used for primarily decorative effects.
== Accela ==
In [[Layer 02]], there is a scene where a narrator explains the workings of [[accela]] while digital graphics are displayed. During this scene, small, nearly transparent blocks of text scroll up the screen. Subject-wise, it has little to no relation with the scene it appears in. It comes from two distinct sources and reads as follows, with line breaks and typographical errors for the most part preserved:
Anti-VEGF
Humanized Monoclonal Antibody<br>
The anti-VEGF antibody is an inhibitor
of angiogenesis (blood-vessel growth)
that may hinder the growth of cancer tumors
by starving their blood supply. Genentech is
investigating this antibody in Phase II<br><br><br>
Vascular endothelial growth factor
(VEGF) is a natural protein that promotes angiogenesis
(blood vessel growth).
VEGF couldpotentially benefit patients who have a
heart that is functioning but has a blocked blood
supply due to artherioscleroticcoronar
The above excerpt is from the web site of Genentech circa 1998.[http://web.archive.org/web/19990222022456/http://www.gene.com/Pipeline/pipeline.html]
Smart materials would be made of nanomachines,
typically microscopic -- with features any size,
down to atomic dimensions. Such machines would
have more or less, the same components as macro,
or familiar "normal" sized machines with
recognizable gears,
bearings, motors, levers and belts... (except for all the nanocomputers).<br>
This is somewhat helpful to the engineer
designing smart materials with a myriad of
functions like shape changing and distributing
fluids and gas -- say for environmental control in
a paper thin space suit that
actively moves with the body or Drexler's smart paint.
Open a can and splat some on a wall. The paint
spreads itself across the surface using microscopic
machines and changes color on command or
becomes a wall sized 3-D television... Then again,
the whole wall may as well be smart material changing
texture or windows on command.<br>
The point here... one can visualize the machines
needed to do such a job: little tractors with sticky
wheels, connection struts and cables to other
machines. Actually, most of this can be done today, only
on a much larger scale and at great expense
(this is where the novel economics of self replicating
machines plugs in). The transition for an engineer,
is using more machines with much smaller parts and
the luxury of vast computing power.
These differences yield more great utility.<br>
Gears made of Buckytubes are great
nanomachine components... Buckytubes are carbon graphite
sheets rolled into a tube (looks like tubes of chicken wire),
and are "like" carbon in its diamond form,
but with ALL available bonding strength aligned on one axis.
These tubes are stronger than diamond
fiber, and the strongest fiber possible with matter,
so we're starting out with real racehorse material.
Globus and Team designs are chemically stable,
very tough and varied in geometry, including gears mad
from "nested" Buckytubes or tubs inside of
tubes. Such a gear would be stiffer and suited for a "long"
drive shaft. And talk about performance...
It's difficult to determine the original source for this, or where it was published, but it appears to be from an article called "Nanotechnology: Magic of Century 21st," a kind of introduction to nanotechnology. [http://www.gubing.com/wbl/Docs/wf/nanotech.doc]
There is also a background image in this scene that originated from a site called Urban Diary. [http://www.cjas.org/~leng/readlist.htm#update]
== KIDS ==
During the scene in [[Layer 06]] where [[Professor Hodgson]] explains [[KIDS]] to Lain, the following scrolls by in very small and difficult-to-read print:
Our initial body of utterances was collected
with a program that periodically called staff
members and asked them to say 5
names selected at random from 64 full
Japanese names (surname followed by first name).
Using this program, 684 utterances
were recorded from 47 native Japanese speakers
(3/4 of which were male) and tagged with the
utterance transcription.
The utterances were represented as Bark-scale
power spectra of 20 ms speech frames, Hamming
windowed at 5 ms shifts. The
utterances were time synchronously phoneme
labeled using their transcriptions in an
automated process. The results were
manually checked and adjusted to correct
any missegmentations.<br>
From this data we generated our initial models as
described above and used them to bring the automated
attendant system
online. The system, open to about 100 users,
ran as described in section 3,
and after some months we had collected over 350
additional utterances. The newly collected
utterances were briefly checked and a few
mislabeled ones were deleted.<br>
Even with the new utterances, this is not
a large data set (especially considering that
the task is multi-speaker, and recorded
over telephone lines), but we nonetheless
performed the following experiments to assess
the effects of incremental retraining.
The 350 new utterances were added in 4 stages
(preserving their temporal sequence) to the
initial set of 684 (e.g. 684+87,
684+175, ...). At each stage one third of
all the utterances were selected at random and
held out for testing. The remaining
two thirds became the training data,
from which a new set of models was made using
the three step procedure outlined
above. <br>
At each stage we made 2 tests. The first checked basic
recognition accuracy when new models were generated from the
expanded training data and the new testing data was
incorporated into the test set. The second used the new testing data
but no new training data in order to check how well
the original models generalized to unseen data.
These two tests were
conducted on both the models produced by embedded
k-means clustering (step 2 above) and on the
models after minimum
error training (step 3 above).
Results for these tests are shown in Figure 2.
The above is an excerpt from a journal article on the development of a computerized voice recognition telephone operator. [http://www.researchgate.net/publication/232642346_A_telephone-based_directory_assistance_system_adaptively_trained_using_minimum_classification_errorgeneralized_probabilistic_descent]
== Infornography ==
In [[Layer 11]] there's more, but i'm tired and i want to go to bed so read about it [http://www.cjas.org/~leng/readlist.htm#update2 here] for now.
For the purposes of this article, typographics includes any situation where text is used for primarily decorative effects.
== Accela ==
In [[Layer 02]], there is a scene where a narrator explains the workings of [[accela]] while digital graphics are displayed. During this scene, small, nearly transparent blocks of text scroll up the screen. Subject-wise, it has little to no relation with the scene it appears in. It comes from two distinct sources and reads as follows, with line breaks and typographical errors for the most part preserved:
Anti-VEGF
Humanized Monoclonal Antibody<br>
The anti-VEGF antibody is an inhibitor
of angiogenesis (blood-vessel growth)
that may hinder the growth of cancer tumors
by starving their blood supply. Genentech is
investigating this antibody in Phase II<br><br><br>
Vascular endothelial growth factor
(VEGF) is a natural protein that promotes angiogenesis
(blood vessel growth).
VEGF couldpotentially benefit patients who have a
heart that is functioning but has a blocked blood
supply due to artherioscleroticcoronar
The above excerpt is from the web site of Genentech circa 1998.[http://web.archive.org/web/19990222022456/http://www.gene.com/Pipeline/pipeline.html]
Smart materials would be made of nanomachines,
typically microscopic -- with features any size,
down to atomic dimensions. Such machines would
have more or less, the same components as macro,
or familiar "normal" sized machines with
recognizable gears,
bearings, motors, levers and belts... (except for all the nanocomputers).<br>
This is somewhat helpful to the engineer
designing smart materials with a myriad of
functions like shape changing and distributing
fluids and gas -- say for environmental control in
a paper thin space suit that
actively moves with the body or Drexler's smart paint.
Open a can and splat some on a wall. The paint
spreads itself across the surface using microscopic
machines and changes color on command or
becomes a wall sized 3-D television... Then again,
the whole wall may as well be smart material changing
texture or windows on command.<br>
The point here... one can visualize the machines
needed to do such a job: little tractors with sticky
wheels, connection struts and cables to other
machines. Actually, most of this can be done today, only
on a much larger scale and at great expense
(this is where the novel economics of self replicating
machines plugs in). The transition for an engineer,
is using more machines with much smaller parts and
the luxury of vast computing power.
These differences yield more great utility.<br>
Gears made of Buckytubes are great
nanomachine components... Buckytubes are carbon graphite
sheets rolled into a tube (looks like tubes of chicken wire),
and are "like" carbon in its diamond form,
but with ALL available bonding strength aligned on one axis.
These tubes are stronger than diamond
fiber, and the strongest fiber possible with matter,
so we're starting out with real racehorse material.
Globus and Team designs are chemically stable,
very tough and varied in geometry, including gears mad
from "nested" Buckytubes or tubs inside of
tubes. Such a gear would be stiffer and suited for a "long"
drive shaft. And talk about performance...
It's difficult to determine the original source for this, or where it was published, but it appears to be from an article called "Nanotechnology: Magic of Century 21st," a kind of introduction to nanotechnology. [http://www.gubing.com/wbl/Docs/wf/nanotech.doc]
There is also a background image in this scene that originated from a site called Urban Diary. [http://www.cjas.org/~leng/readlist.htm#update]
== KIDS ==
During the scene in [[Layer 06]] where [[Professor Hodgson]] explains [[KIDS]] to Lain, the following scrolls by in very small and difficult-to-read print:
Our initial body of utterances was collected
with a program that periodically called staff
members and asked them to say 5
names selected at random from 64 full
Japanese names (surname followed by first name).
Using this program, 684 utterances
were recorded from 47 native Japanese speakers
(3/4 of which were male) and tagged with the
utterance transcription.
The utterances were represented as Bark-scale
power spectra of 20 ms speech frames, Hamming
windowed at 5 ms shifts. The
utterances were time synchronously phoneme
labeled using their transcriptions in an
automated process. The results were
manually checked and adjusted to correct
any missegmentations.<br>
From this data we generated our initial models as
described above and used them to bring the automated
attendant system
online. The system, open to about 100 users,
ran as described in section 3,
and after some months we had collected over 350
additional utterances. The newly collected
utterances were briefly checked and a few
mislabeled ones were deleted.<br>
Even with the new utterances, this is not
a large data set (especially considering that
the task is multi-speaker, and recorded
over telephone lines), but we nonetheless
performed the following experiments to assess
the effects of incremental retraining.
The 350 new utterances were added in 4 stages
(preserving their temporal sequence) to the
initial set of 684 (e.g. 684+87,
684+175, ...). At each stage one third of
all the utterances were selected at random and
held out for testing. The remaining
two thirds became the training data,
from which a new set of models was made using
the three step procedure outlined
above. <br>
At each stage we made 2 tests. The first checked basic
recognition accuracy when new models were generated from the
expanded training data and the new testing data was
incorporated into the test set. The second used the new testing data
but no new training data in order to check how well
the original models generalized to unseen data.
These two tests were
conducted on both the models produced by embedded
k-means clustering (step 2 above) and on the
models after minimum
error training (step 3 above).
Results for these tests are shown in Figure 2.
The above is an excerpt from a journal article on the development of a computerized voice recognition telephone operator. [http://www.researchgate.net/publication/232642346_A_telephone-based_directory_assistance_system_adaptively_trained_using_minimum_classification_errorgeneralized_probabilistic_descent]
== Infornography ==
In [[Layer 11]] there's more, but i'm tired and i want to go to bed so read about it [http://www.cjas.org/~leng/readlist.htm#update2 here] for now.
