Big data or will be able to contribute to the world artist

Nowadays, big data has been applied more and more widely in more and more fields, and its contribution to the world is also increasing. In fact, big data is not just about the size of the data. It is equally important to discover and utilize different sources of data. These data, including web data, sensor data, location data, etc. These are generally known to us. Now, add the art data. And it is not the data of masterpieces of the world, but the data of one stroke and one stroke drawn when we draw things. What will happen?

At this week's SIGGRAPH meeting in Los Angeles, there are two statements about the study. Their research shows that if the data that can be collected are correct, a comprehensive understanding of how one draws can be made. One of the projects was done by Carnegie Mellon University and the Microsoft Research team to get the subject to play a game on the iPhone between Pictionary and Wheel of Fortune in order to get the data. This research was especially interesting from the point of view of generating data only with crowdsourcing, taking advantage of the ubiquity of the iPhone and finally collecting 17,000 drawings for a data set.

They asked the subject to play a game called DrawAFriend and draw their finger to some celebrity or friend's picture of each other. After a participant completes, the drawn pictures are presented one after the other to the other participant, and the other participant guesses who the painting is based on to their own certainty. The purpose is to compare the strokes of each participant, the less the number of guesses, the higher the stroke score in the painting.

How to play DrawAFriend

The researchers then used a number of corrections to correct some of the bad strokes and masturbation fingers that the user did not know. What these researchers need to do next is to look for different ways to analyze the user's behavior with collected data (such as when "cancel" functions, stroke order, etc. are used) and they are also thinking about how strokes can be made more accurate Beauty can look more beautiful.

Another study from Disney. They looked for some art staff and asked them to draw a sketch of 24 different people using a stylus. Each of the 24 people needed to draw four times, each time less than the previous time (270, 90, 30 and 15 seconds). In this way, researchers collected 8,000 strokes for each artist.

The researchers said that data analysis can determine the difference between the artist's sketch and the original in terms of geometric features (such as close eyes or too chin outline) and can help them find bad habits that need to be corrected.

An example of sketch and analysis at Disney Research

This kind of research is relatively new, indicating that it is possible to collect and analyze almost any type of digital data, but I can imagine someone asking what is the use. Maybe a company like Disney can use this technology as a programming engine to mass-produce a sequel to popular TV shows or to build educational software to help ambitious artists overcome their own illnesses.

Otherwise, it seems that an artist's work is interesting because of its uniqueness. If there's an algorithm that corrects and optimizes strokes, does that work get me done? And, as the Disney researchers point out, some variants are sometimes more about each person's style than Not caused by inaccurate copying.

It is important that we have the ability to collect and analyze such data and to achieve this (as we did with Carnegie Mellon University research projects). I'm sure someone will find many applications for these art-based algorithms, but I do not know it. However, it seems that sometimes it is not necessary for us to quantify something in this world simply because we have this ability - art can be one of them.