Big Data new uses: creating excellent artists

Big data is not just about the size of the data: It's also important to find and use sources of different data. Data can be Web page data, sensor data, location data, and the current art data. No, it's not the data from the World Masters, it's about the amount of data we use to draw things.

At the SIGGRAPH meeting in Los Angeles this week, there were two speeches about research. Their http://www.aliyun.com/zixun/aggregation/6551.html "> research results show that if the data collected is right, you can have a comprehensive understanding of how a person paints. One of the projects was done by Carnegie Mellon University and the Microsoft Research team, who, in order to get the data, let the subjects play a game between Pictionary and Wheel of Fortune on the iphone. The research is particularly interesting in terms of generating data from crowdsourcing methods, using the iphone's ubiquity, and finally collecting 17,000 sketches for a data collection.

They asked the subjects to play a game called drawafriend, using their fingers to draw pictures of celebrities or friends who knew each other. After a participant finishes, the drawing is then presented to another participant in one stroke, and the other participant guesses the person who drew it according to the degree of certainty. The aim 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 some corrective methods to correct the bad strokes and the fat fingers that the user did not know. The next step for the researchers is to find different ways to analyze user behavior by collecting data (such as when to use the "Cancel" function, the order of strokes, etc.), and they are also considering how to make strokes more accurate and aesthetically pleasing.

Another study comes from Disney. They found some art workers and asked them to sketch 24 different people with a stylus, and each of the 24 had to be painted 4 times, each time less than the previous one (270, 90, 30, and 15 seconds). In this way, the researchers collected 8,000 strokes for each art worker.

The researchers say data analysis can determine the difference between the artist's sketch and the original, such as the eye is too close or the chin contour is too thick, and can help them find a bad habit to correct.

A sketch and analysis example from Disney research

This kind of research is quite novel, which means that it is possible to collect and analyze almost any type of digital data, but I can also imagine that someone would ask what it would be. Maybe a company like Disney could use this technology to make a program engine that could produce a video sequel to a popular movie or use it as an educational software to help aspiring artists overcome their bad habits.

Otherwise, the work of an artist seems to be because of its uniqueness. If there is an algorithm that can correct and optimize strokes, the resulting work is not what I finished it? And, as Disney researchers point out, some variants are sometimes more of a person's style than a result of inaccurate replication.

It is important that we have the ability to collect and analyse such data and reach such a scale (like Carnegie Mellon University research Project). I'm sure someone will find a lot of apps for these art algorithms, but I don't know. But there seems to be times when we don't need to quantify something in this world simply because we have the ability-art can be said to be one of them.