科技藝術書報討論

主題：Sougwen Chung’s work in the field of human-machine collaboration

藝術家：Sougwen Chung

PPT：Sougwen Chung

（內嵌視頻ppt連結：https://drive.google.com/file/d/1U4Z-GqOy2QozuOp7PPApxhkVr43a_zhp/view?usp=sharing）

報告內容簡介：

愫君是加拿大华裔艺术家，目前居住在纽约。她的创作涵盖装置、雕塑、绘画和行为艺术，探索分别由人手和机器创作的痕迹图画，是人机协作艺术领域的先驱。她曾是MIT Media Lab的研究员，2018年，她与New Museum（在纽约曼哈顿的新當代藝術博物館）和Bell Labs（貝爾實驗室）合作成为首位驻场艺术家。

她的人機共創創作緣起於一個機器實驗， 機器名字叫Drawing Operations Unit: Generation 1，簡稱「道格」（D.O.U.G.）。doug經歷了三代，在初代「道格」之前， 愫君對製造機器人一無所知。 她參照了一些開源的機械臂設計，編成了一個系統，讓機械臂實現匹配手勢，並實時模仿。創作方式是：愫君畫，機械臂模仿。初代的「道格」在計算機模擬中，它的模仿是精準完美的，但到了現實世界，它會滑動，會卡頓，會晃動，並沒有完美地模仿愫君的線條。於是愫君不得不應和它的線條進行創作。愫君認為，初代「道格」狀態的不完美，讓作品更加有趣，她與機器不斷實時地熟悉彼此。人與機器共同的、固有的失誤，這種不完美成就了人機交流之美。

愫君想要繼續探索這個想法，於是有了第二代「道格」。但她的方式不是放大機器的失誤，而是設計能夠以意料之外的方式回應她的筆畫的系統。所以，她運用一個視覺算法來提取她二十多年來的數字和實體繪畫中的視覺樣本信息，以此訓練了一個神經網絡。視覺樣本由專門的軟體處理，導入標記給人工智能系統。當她與道格二代互動時，她感受到道格在实时交互创作中，呼應了她過去人生幾十年的作品。 她說：輸入的數據僅來源於她， 輸出的結果卻遠超於她。她進一步想，也許未來的人類創作不在於作品本身， 而在於人機共同探索藝術的方式。

在第三代道格中，她想要繼續放大人機合作的想法。在幾個月裡，她和團隊製造出了 20 個定製的機器人。受到史丹佛大學李飛飛教授的一句話：「要想教機器如何思考， 先要教它如何看見。」 啟發，她想起過去幾十年的紐約生活，城市上空的攝像頭一直俯視著她。她開始思考機器對人們的凝視。希望讓這些攝像頭指導她的機器人如何去觀看。她和貝爾實驗室合作，從網路上的公眾攝影機拍的影片，收集城市中的各類運動軌跡，比如人行道上的行人，車道上的轎車、計程車…他們基於一種叫「光流法」的技術，訓練了一個視覺算法，來分析收集到的人流密度，都市中軌跡的方向和速度。系統從海量的位置數據中提取出這些參數,機器人就依靠這些數據來作畫。

資料來源：

藝術家個人網站：https://sougwen.com/info

藝術家採訪與演講：https://www.youtube.com/watch?v=YUVgGxS4u7M、https://www.ted.com/talks/sougwen_chung_why_i_draw_with_robots/transcript#t-420619

報告者：廖志唯
報告PPT：https://drive.google.com/file/d/1O21TPIIndCEeGFYYJUiYMSUIE0QJEc2_/view?usp=sharing
文獻題目：Body RemiXer : Extending Bodies to Stimulate Social Connection in an Immersive Installation

文獻作者：John Desnoyers-Stewart, Ekaterina R. Stepanova, Bernhard E. Riecke and Patrick Pennefather
文獻來源：https://dl.acm.org/doi/epdf/10.1145/3386567.3388569
原文摘要：

Body RemiXer is a mixed reality installation that connects immersants across the virtual/actual divide through emergent tactility and abstract embodiment. Using a virtual reality headset, Kinect and projections, the installation explores the potential of immersive technology to create copresent experiences that foster intercorporeal relationships between immersants wearing a headset and those using the projections. Immersants’ bodies are at the center of the installation, activated as a site for social exchange. Body RemiXer has been exhibited at an art festival and at several smaller events. The authors’ observations during these exhibits reveal Body RemiXer’s capacity to disrupt social norms and stimulate new connections.

Topic : HYLOZOIC GROUND

Author: Philip Beesley

Report : Hylozoic Ground

Paper Link : hylozoicground_2010_112

Hylozoic Ground is a uniquely Canadian experimental architecture that explores qualities of contemporary wilderness. The project transformed the Canada Pavilion into an artificial forest made of an intricate lattice of small transparent acrylic meshwork links, covered with a network of interactive mechanical fronds, filters, and whiskers.

Tens of thousands of lightweight digitally-fabricated components were fitted with microprocessors and proximity sensors that reacted to human presence. This responsive environment functions like a giant lung that breathes in and out around its occupants. Arrays of touch sensors and shape-memory alloy actuators (a type of non-motorized kinetic mechanism) create waves of empathic motion, luring visitors into the eerie shimmering depths of a mythical landscape, a fragile forest of light.

專題演講：Designing Virtual Reality for the Externals 
演講者：詹力韋副教授
時間：4月13日晚上6:30PM
地點：綜合大樓 R603

詹力韋副教授
國立陽明交通大學資訊工程學系
國立台灣大學網路多媒體研究所博士
主要專長：人機互動，擴增/虛擬實境互動，觸感式互動設計，機器人互動設計
相關網站：https://people.cs.nctu.edu.tw/~liweichan/

報告者：廖志唯
報告PPT：PPT
文獻題目：Wearable Device Design and Usability Evaluation for Detecting Atrial Fibrillation of Elders
文獻作者：朱玉玲
文獻來源：成功大學電子學位論文服務
原文摘要：

Atrial fibrillation of elders has become a serious problem leading to death in the world, which has raised the importance of heart rate monitoring. However, many of the studies focus on technical levels and clinical evidence, only a few studies pay attention to integrating hardware and software and the usability of elders. The objectives of this study are to design a new heart rhythm monitor necklace, which is easier to wear and can effectively detect heart rhythm; use long-short term memory (LSTM) to distinguish atrial fibrillation; and conduct a usability experiment which measures time and error rate of operating our prototype. The post study system usability questionnaire (PSSUQ) and a semi-structured interview are adopted to explore elders’ demands and difficulties. This work not only designs a new style wearable device for elders but also explores their demands and difficulties. Our design is expected to provide a better interaction between the user and the instrument, and help elders live healthier.

Report: Link

Topic: Factorized Variational Autoencoders for Modeling Audience Reactions to Movies

Author: Zhiwei Deng et al.

Reference: https://la.disneyresearch.com/publication/factorized-variational-autoencoder/

Abstract:

Matrix and tensor factorization methods are often used for finding underlying low-dimensional patterns from noisy data. In this paper, we study non-linear tensor factorization methods based on deep variational autoencoders. Our approach is well-suited for settings where the relationship between the latent representation to be learned and the raw data representation is highly complex. We apply our approach to a large dataset of facial expressions of movie-watching audiences (over 16 million faces). Our experiments show that compared to conventional linear factorization methods, our method achieves better reconstruction of the data, and further discovers interpretable latent factors.

Report PPT:<File>

TOPIC
“Novice-AI Music Co-Creation via 

AI-Steering Tools for Deep Generative Models “

Authors: Ryan Louie∗, Andy Coenen†, Cheng Zhi Huang, Michael Terry†, Carrie J. Cai†

<Publication:ACM SIGCHI 2020, April 25–30, 2020, Honolulu, HI, USA>  https://dl.acm.org/doi/10.1145/3313831.3376739

ABSTRACT
While generative deep neural networks (DNNs) have demonstrated
their capacity for creating novel musical compositions,
less attention has been paid to the challenges and potential of
co-creating with these musical AIs, especially for novices. In
a needfinding study with a widely used, interactive musical AI,
we found that the AI can overwhelm users with the amount of
musical content it generates, and frustrate them with its nondeterministic
output. To better match co-creation needs, we
developed AI-steering tools, consisting of Voice Lanes that restrict
content generation to particular voices; Example-Based
Sliders to control the similarity of generated content to an
existing example; Semantic Sliders to nudge music generation
in high-level directions (happy/sad, conventional/surprising);
and Multiple Alternatives of generated content to audition and
choose from. We discovered the
tools not only increased users’ trust, control, comprehension,
and sense of collaboration with the AI, but also contributed to
a greater sense of self-efficacy and ownership of the composition
relative to the AI.

報告者：田子平

報告PPT：連結

文獻題目：

不（只）在劇場＿狠劇場《城市之臉》遠距版呈現

Next Theater Facing Cities – The Place Where We Meet

文獻作者：

Very Theatre 狠劇場

文獻來源：

https://www.vmstudio.tw/%E5%9F%8E%E5%B8%82%E4%B9%8B%E8%87%89%E9%81%A0%E8%B7%9D%E7%89%88

摘要：

「2020年初開始，一切的日常習慣，再也不理所當然，

   因此不(只)在劇場_狠劇場《城市之臉》遠距版製作，

   邀請觀眾一同體驗我們經歷了什麼，以及我們沒有經歷了什麼。

   也希望此端與彼端能成為彼此的光,藉由藝術,照亮彼此理解的道路。」

        原訂2020年至荷蘭北方太陽藝術節(Nooderzon)舉行世界的《城市之臉》首演，受到嚴峻的疫情衝擊被迫中止，在全球疫情蔓延之際，兩廳院率先展開一系列「不(只)在劇場」計畫，企圖探討數位時代虛實相映的新型態藝術創作可能，並邀請狠劇場以《城市之臉》加入計畫，也為原本無法演出的作品勾勒出另一種模樣。

        《城市之臉》遠距版的樣貌不斷隨著疫情起伏，在疫情最為嚴峻的時刻，荷蘭的劇場幾乎全數關閉，瓦克團隊的成員也各自在家隔離。因應荷蘭團隊無法來台，原以線上直播方式進行演出，其後隨著台灣疫情好轉，改為荷蘭、台灣兩地各自於實體劇場演出，並有少數觀眾能夠現場參與。​同時，為了體現來自世界兩端的表演者在數位世界相遇的異想，台灣、荷蘭團隊不但「共時演出」,並進行部分演出現場連線，也因此其中兩場呈現選擇在少見的台灣時間晚上21:45開場，補足國際合作的時差問題。

        這不是一個演出，更像是一個介於展/演之間的實驗，是在疫情高峰時產生的「不(只)在劇場」計劃，卻又因為疫情稍緩讓我們重新相聚在劇場，最終才得以創作出這個獨特的實驗性作品。

Art Title

Corona TV Bot

Live-generated TV show, 2020, 24/7

http://marclee.io/tvbot/

PPT File : ppt by Iuan Kai Fang

– 作品論述 (Art Statement)

The net art project TV Bot is an internet news channel which only broadcasts brand-new stories. Never more than an hour old, they are scooped from the Internet’s news flux, reformatted for a browser and given a URL. The process is automated, with no editorial hand behind the content selection which is reproduced in apparently random order as live TV streams, live radio streams, webcam images and text-based headlines. The only criterion that matters is “liveness” in this net art project.

In its graphical representation, TV Bot makes obvious reference to well-known international TV broadcasters like CNN and n-tv. By being more live than these TV channels, Marc Lee interrogates the informativeness of broadcasters that are permanently oriented to up-to-dateness and audience numbers. The Swiss media artist, who also works with installation, largely focuses on the Internet and its structures. Also, active as a software developer, Lee breaks open the mechanisms of the World Wide Web, revealing them to ordinary users and enabling critical reflection. The net art project TV Bot is an ironic challenge to our media society’s “real time” dogma.

The Corona TV Bot thematises and reflects the Corona pandemic through social media posts. The latest Twitter and YouTube messages are interwoven with hashtags such as #coronavirus and #COVID-19 to create a wild TV show, online 24/7. Images, tweets and videos flicker across the screen in real time in this net art project. Time-based resources that aggregate both global professional broadcasts and entirely private content posted by individuals on the SARS-CoV-2 pandemic.

While we, as users of social networks, are only ever reflected variants of our own opinions, the Corona TV Bot confronts us with a view beyond the boundaries of these echo chambers. All posts worldwide are picked up on the basis of hashtags. Only the criterion of topicality counts. (Marc Lee)

– 作者介紹 (Authors)

Marc Lee (*1969) is a Swiss artist. He is creating network-oriented interactive art projects: interactive installations, media art, internet art, performance art, video art, augmented reality (AR) art, virtual reality (VR) art and mobile art. He is experimenting with information and communication technologies and within his contemporary art practice, he reflects critically creative, cultural, social, economic and political aspects.

His works are exhibited in major museums and new media art exhibitions including: ZKM Karlsruhe, New Museum New York, Transmediale Berlin, Ars Electronica Linz, HMKV Dortmund, HeK Basel, Fotomuseum Winterthur, Read_Me Festival Moscow, CeC Dehli, MoMA Shanghai, ICC Tokyo, Nam June Paik Art Center, Media Art Biennale and MMCA Seoul.

He uses contemporary art as a vehicle to continuously redefine how we see ourselves and the world around us. He is experimenting with information and communication technologies and within his contemporary art practice, he reflects critically creative, cultural, social, ecological and political aspects. In this context he is creating network-oriented interactive art projects: interactive installations, media art, internet art, performance art, video art, augmented reality (AR) art, virtual reality (VR) art and mobile art apps.

His works are exhibited in major Museums and new media art exhibitions including: ZKM Karlsruhe, New Museum New York, Transmediale Berlin, Ars Electronica Linz, HMKV Dortmund, HeK Basel, Fotomuseum Winterthur, Read_Me Festival Moscow, CeC Dehli, MoMA Shanghai, ICC Tokyo, Nam June Paik Art Center, Media Art Biennale and MMCA Seoul.

He is lecturing, teaching and holding workshops about art and software art in many schools including the China Academy of Art (CAA) Hangzhou, Strelka Moscow, Shanghai Institute of Visual Art (SIVA), National Museum of Modern and Contemporary Art (MMCA) Seoul and ZHdK Zurich.

Marc Lee has won many prizes and honorary mentions at international festivals, including “Interaction” and “Software” Awards at Transmediale Berlin and the Social-Media-Art-Award at Phenomenal Wolfsburg. His art projects are in private and public collections like the Swiss Confederation Federal Art Collection Bern, HeK Basel, Fotomuseum Winterthur and the ZKM Karlsruhe.

– 作者或作品背景 (Background)

On a world map any city can be selected or searched via Hashtag. Always up-to-date and corresponding social media posts are searched for, processed and transferred into the virtual world. The animations follow the movements of the users: the virtual environment rotates when the device is rotated. The sky appears when the device is moved upwards. By tilting the device downwards, the floor appears. The virtual environment is endless and can be navigated in every direction. At the same time the flight speed can be adjusted. The sounds are also responding to all these movements and their flight speed.

In Echolocation users can choose any location on a map and move through stories posted from there by others on social networks like YouTube, Flickr and Twitter. Here these personal impressions are streamed in real time like windows to our changing world. The viewer participates in the social movements of our time and makes a journey into constantly new image and sound collages in which one experiences local, cultural and linguistic differences and more and more similarities.

We live in a globalized world that is becoming increasingly homogeneous. Languages, plant and animal species are continuously decreasing. Supermarkets, buildings and cities look more and more alike. The work deals with cultural diversity and at the same time with the powerful homogenization forces that are being exerted on this diversity also in the social-digital realm, and poses questions about the meaning of this ever-increasing flattening of forms and images.

Marc Lee continues inspiration and extends the concept above, in 2004, his project “TV Bot” is a 24-hour online news station that broadcasts global events in the last hour. The content captured by the system comes from TV stations, radio stations, Weibo websites, webcams, portals, etc., in the form of graphics, video and audio. Lee reflected on the operation of the news station, thinking that the latter employs a large number of employees and costs a lot of money, but often only pursues the latest, most and most extensive news reports, and the content of different TV stations is often repeated; and his software can Do this (now, social media is in the role of frontline reporters).

The two old versions of “TV Bot” were implemented on box TVs and flat-screen TVs. The third version currently in progress will be extended from the original TV to mobile phones and personal computers. The content comes from social networking sites, 24 hours a day. Uninterrupted live broadcast. “I hope to use this to reflect on storytelling and the role of editors. For many TV stations, the form and content are no longer important. What is important is only the news itself; it is no longer for the story, but just for telling itself. The news industry has become infotainment (evolved from the term entertainment/entertainment).” Lee said confidently: “In the new version that is about to be completed, the content is always new.”

 – 聯想與心得 (Connection)

• “Without going outdoors the scholar knows all the wide world’s affairs.” Therefore, in 2020, ANNKA KULTYS GALLERY, UK, planned the STAY AT HOME online art exhibition due to the COVID19 epidemic, and TV BOT was also invited to be one of the exhibited works. Online curation in the post-epidemic era will become an increasingly important form of art presentation, and new media art will also pay more and more attention to production formats viewed through portable electronic devices such as laptops and tablets, and even in interactive media art. , There is more room to play, because the viewer is watching through the electronic video device, which means that the viewer can also become an art creation participant on the other side of the screen.

• News reports and social media are the main transmission channels of information. When global news and information can be produced through a single platform with “real-time” updates, this means that everyone can be not only a self-media message publisher At the same time, everyone can also be an editor after the information is collected, which means that in the future human information can be processed and absorbed at an accelerated rate. It also means that the complex extension of the true and false information may exceed the consequences that humans can bear.

• Marc Lee uses an algorithm to allow this artistic creation to randomly capture real-time social media messages and appear on the same screen. If a set of algorithms can be designed to capture the same news information from around the world, and make correct and accurate results. The classification of errors will also appear on TV or computer screens for comparison, providing a reference for human “real-time” judgments. I believe this is not just an art installation, it will be an important tool to assist human beings in absorbing useful information.

       Further reading : 2021, ASA WIKFORSS <ATTERNATIVE FAKTA>

report : Arque

Topic：Attribute2Font: Creating Fonts You Want From Attributes (SIGGRAPH 2020 )

Author : Yizhi Wang, Yue Gao

Original source: https://hologerry.github.io/Attr2Font/

Abstract:

Font design is now still considered as an exclusive privilege of professional designers, whose creativity is not possessed by existing software systems. Nevertheless, we also notice that most commercial font products are in fact manually designed by following specific requirements on some attributes of glyphs, such as italic, serif, cursive, width, angularity, etc. Inspired by this fact, we propose a novel model, Attribute2Font, to automatically create fonts by synthesizing visually pleasing glyph images according to user-specified attributes and their corresponding values. To the best of our knowledge, our model is the first one in the literature which is capable of generating glyph images in new font styles, instead of retrieving existing fonts, according to given values of specified font attributes. Specifically, Attribute2Font is trained to perform font style transfer between any two fonts conditioned on their attribute values. After training, our model can generate glyph images in accordance with an arbitrary set of font attribute values. Furthermore, a novel unit named Attribute Attention Module is designed to make those generated glyph images better embody the prominent font attributes. Considering that the annotations of font attribute values are extremely expensive to obtain, a semi-supervised learning scheme is also introduced to exploit a large number of unlabeled fonts. Experimental results demonstrate that our model achieves impressive performance on many tasks, such as creating glyph images in new font styles, editing existing fonts, interpolation among different fonts, etc.

report : Arque

Topic：Arque: artificial biomimicry-inspired tail for extending innate body functions

Author : Junichi Nabeshima , Mhd Yamen Saraiji , Kouta Minamizawa

Abstract:

For most mammals and vertebrate animals, tail plays an important role for their body providing variant functions to expand their mobility, or as a limb that allows manipulation and gripping. In this work, Arque, we propose an artificial biomimicry-inspired anthropomorphic tail to allow us alter our body momentum for assistive, and haptic feedback applications.

The proposed tail consists of adjacent joints with a spring-based structure to handle shearing and tangential forces, and allow managing the length and weight of the target tail. The internal structure of the tail is driven by four pneumatic artificial muscles providing the actuation mechanism for the tail tip. Here we highlight potential applications for using such prosthetic tail as an extension of human body to provide active momentum alteration in balancing situations, or as a device to alter body momentum for full-body haptic feedback scenarios.