We found an Airbnb security issue and made a game with it.

‘Airbnb Hosts’ is a browser game using Airbnb data where you need to guess the owner of the room. The game presents you a picture of an Airbnb room and 3 Airbnb Hosts. You need to match the owner with the room!

You can play the game with rooms of your favourite city. All the data is located on Airbnb servers.

We are not sponsored by Airbnb but we added a button “Book This room” so Airbnb will not to angry using their data and have it online.

This paper presents a simple method for "do as I do" motion transfer: given a source video of a person dancing we can transfer that performance to a novel (amateur) target after only a few minutes of the target subject performing standard moves. We pose this problem as a per-frame image-to-image translation with spatio-temporal smoothing. Using pose detections as an intermediate representation between source and target, we learn a mapping from pose images to a target subject's appearance. We adapt this setup for temporally coherent video generation including realistic face synthesis.

We study the problem of video-to-video synthesis, whose goal is to learn a mapping function from an input source video (e.g., a sequence of semantic segmentation masks) to an output photorealistic video that precisely depicts the content of the source video.

While its image counterpart, the image-to-image synthesis problem, is a popular topic, the video-to-video synthesis problem is less explored in the literature. Without understanding temporal dynamics, directly applying existing image synthesis approaches to an input video often results in temporally incoherent videos of low visual quality.

In this paper, we propose a novel video-to-video synthesis approach under the generative adversarial learning framework. Through carefully-designed generator and discriminator architectures, coupled with a spatial-temporal adversarial objective, we achieve high-resolution, photorealistic, temporally coherent video results on a diverse set of input formats including segmentation masks, sketches, and poses.

Experiments on multiple benchmarks show the advantage of our method compared to strong baselines.  In particular, our model is capable of synthesizing 2K resolution videos of street scenes up to 30 seconds long, which significantly advances the state-of-the-art of video synthesis. Finally, we apply our approach to future video prediction, outperforming several state-of-the-art competing systems.

Our ongoing series ‘Inventory’ is a cultural project where we dig the ephemeral characteristics of the public space and street furniture. Photogrammetry enables us to collect objects in detail and it is possible to create compositions using these objects as units.

We started shooting in our own neighborhood (Kadıköy, Istanbul) practicing on methodology & process of capturing objects in wide range, complexity, size and number. As we traveled for other occasions we’ve always captured in order to expand our inventory: We shot on the streets of Manhattan for a week capturing 400 unique street objects such as fire hydrants, utility poles, booths and statues. We shot 150 ancient wells-fountains in Venice and 200 facades-sidewalks in Kreuzberg-Berlin

Public space collects culture constantly as the surfaces are painted with tags, graffiti and covered with stickers. Conditions change with weather, impacts and maintenance. Interferences create uniqueness, but unlike cultural objects under protection in museums or galleries, street furniture is under constant danger of renovation or replacement. Using photogrammetry, we are documenting and protecting street culture in 3-dimensions with high-resolution texture.

During the creation of our first composition of Istanbul street objects, we let the coincidences lead the way. A random arrangement was appropriate referencing the chaotic soul of the city. During the composition of Manhattan inventory, we rather used an arrangement similar to grid based city plan with an emphasis on vertical growth of the city. Venice inventory installed around the Grand Canal and Berlin seeks the german precision.

We are working on other stills and videos as well as online-VR presentations.

We introduce a data-driven approach for unsupervised video retargeting that translates content from one domain to another while preserving the style native to a domain, i.e., if contents of John Oliver's speech were to be transferred to Stephen Colbert, then the generated content/speech should be in  Stephen Colbert's style. Our approach combines both spatial and temporal information along with adversarial losses for content translation and style preservation. In this work, we first study the advantages of using spatiotemporal constraints over spatial constraints for effective retargeting.  We then demonstrate the proposed approach for the problems where information in both space and time matters such as face-to-face translation, flower-to-flower, wind and cloud synthesis, sunrise and sunset. 

Many motile microorganisms react to environmental light cues with a variety of motility responses guiding cells towards better conditions for survival and growth. The use of spatial light modulators could help to elucidate the mechanisms of photo-movements while, at the same time, providing an efficient strategy to achieve spatial and temporal control of cell concentration. Here we demonstrate that millions of bacteria, genetically modified to swim smoothly with a light controllable speed, can be arranged into complex and reconfigurable density patterns using a digital light projector. We show that a homogeneous sea of freely swimming bacteria can be made to morph between complex shapes. We model non-local effects arising from memory in light response and show how these can be mitigated by a feedback control strategy resulting in the detailed reproduction of grayscale density images. 

A simple implementation of the pix2pix paper on the browser using TensorFlow.js. The code runs in real time after you draw some edges. Make sure you run the model in your laptop as mobile devices cannot handle the current models. Use the mouse to draw. 

Neural Beatbox?  RNN-based Rhythm Generation + Audio Classification = FUN!

Built with tensoflow.js, magenta.js and p5xjs. Rhythm generation part was based on magenta’s DrumRNN and teropa ’s Neural Drum Machine!

I use my own keras model (converted to TFJS model) to classify sound segments.

In the XIX century, under the influence of the tuberculosis epidemic, the idea of «body image» was developed and presented for the first time. A new image of aristocratic appearance was created. The romanticization of tuberculosis is the first popular example of a very modern trend — the «construction» of personality as an image. Nowadays the idea of individuality is an essential attribute of any person, and most of us spend a lot of time and effort to create a personal image of our body, a «personal brand». In the future, I believe, this trend will become even more widespread, and the concept of «beauty» will continue to change. These are luminous decorative objects that are designed to be worn in the mouth. They are made of orthodontic plates, medical silicone, electronic components and batteries. Through these devices I am projecting a possible scenario of development of beauty trends in the near future. The geometry of the face and its sensation have been changed. Perhaps it is not “beautiful” in the ordinary sense, but is definitely attractive aesthetically. These devices bring us closer to a time when it will be possible to implant light sources and other electronic components into the body, just “for appearance”, as, for example, piercing nowadays.

SXSW 2018: A talk and live performance to premiere an AI capable of live improvisation, in duet, with a human vocalist.

A post shared by Patrik Hübner Creative Coding (@patrikhuebner) on Aug 2, 2018 at 9:08am PDT

A post shared by Brute wine (@brutewine) on Aug 2, 2018 at 8:08am PDT

BRUTE is a data-driven wine brand that distills the core weather elements of wind, rain & sun into a holistic visual and sensual experience. It uses historic and real-time weather data to shape its identity. So, as each vintage evolves with the climate, so does the brand. 

Analog sound generators, based on magnetic tape and optical components controlled via graphic score composed with digital interface.

The APPARATUM has been inspired by the heritage of the Polish Radio Experimental Studio – one of the first studios in the world producing electroacoustic music.

The installation draws inspirations musically and graphically from the “Symphony – electronic music” – composed by Bogusław Schaeffer

Bogusław Schaeffer conceived his own visual language of symbols that conveyed the cues for the sound engineer responsible for the production of the piece.

The physical form is inspired by the general aesthetics of the Studio’s famous “Black Room” designed by Oskar Hansen. The electroacoustic generators and filters were arranged in a modular fashion inside two steel frames – the construction element that we’ve referred to in our design.

Portable Lumipen is a portable dynamic projection mapping system that tracks a target at 1,000fps with 3ms response time.

In recent years, projection mapping systems for dynamically moving targets have been proposed and this field is called "dynamic projection mapping." In Dynamic projection mapping, the moving target is tracked by a high-speed camera, and the projection is controlled at high speed. By achieving high responsiveness in sensing and feedback, these systems enable dynamic and immersive applications. However, they are not truly dynamic because the systems are fixed in a room.

In this reseach, we use a 3D-stacked vision chip and an optical gaze contoller and realize a portable dynamic projection mapping system. The 3D-stacked vision chip enables to omit a huge workstation and inefficient data transfer, and the optical gaze controller enables high-speed tracking shot and also tracking projection. The portable system enables a variety of applications including wearable user interface, projector-based makeup and combination with a drone, and this device will take projection mapping a step further.