Ai can now recreate famous paintings with a 3D printer


Can artificial intelligence and 3D printer help fight the world's greatest threat?

A team of Computer Computer Science and Artificial Intelligence Laboratory (CSAIL) thinks so.

A new system that has aimed at designing paintings, uses 3D printing and deep learning to "Original Recreate Favorite Paintings" -Regularness of lighting conditions or placement.

"If you just reproduce the color of a painting as it looks at the gallery, it may look different in your home," study co-author Changil Kim, a post-secretary fellow at Xail, said in a statement.

So, instead of using the traditional four fixed inks (Cyan, Magenta, Yellow, Black) found in 2D printers, the researchers used a special technique they called "color-contoning."

The team trained a deep-learning model to predict the optimal stack of different inks (via MIT CSAIL)

The process involves a 3D printer and 10 different transparent inks in their thin layers-acin to the wafers and chocolate of a kit-cut bar.

By combining this method with a decade-old approach to half-toning (which creates a gradient-like effect through dots), they are able to better capture the "nuances of the colors".

Based on experimental replication of various oil paintings created by an artist collective, the team found that repaint was more than four times more accurate at creating accurate color shades than state-of-the-art physical models.

"Our system works under any lighting condition, which shows a much larger color reproduction ability than almost any other previous job," said Kim.

There is one catch, though: CSAIL's facsimiles are just about the size of a business card. After all, 3D printing is not cheap.

In the future, however, they show that advanced business printers will allow for greater paintings, paving the way for a more efficient system.

Still, the question remains: Which ink should be used for what colorings?

As with many tasks in the days, humans have passed the load of selection on a deep-known model, which can predict the optimal stack of hues in each area of ​​the canvas.

Repaint's creators envision its use in remaking artwork for a home, protecting origins in museums, and producing prints of historical pieces.

Research found that more than four times more reappropriate than state-of-the-art physical models at creating the exact color shades for different artworks (by MIT CSAIL)

The program has a way to go, though, before it can start hopping up duplicates from "Starry Night."

For starters, it can not fully reproduce a few colors-like cobalt blue-right to a limited ink library, which mechanical engineer Mike Focie hopes to expand soon.

And, as you can see in the pictures and videos above, there is something defensively lacking in repaid's imitations: texture. The team will continue to work to better detail, eventually hoping to create special effects like glossy and matt finishes.

"The value of fine art has rapidly increased in recent years, so there is an increased tendency for it to be locked up in warehouses, away from the public eye," Fox said. "We build the technology to reverse this trend, and make cheap and accurate reproductions that can be used by all."

A full report was issued this week by MIT CSAIL.

More Coverage on

Source link