Monday , April 19 2021

3ders.org – MIT RePaint uses AI and 3D print to accurately reproduce images for your home



November 29, 2018 | Thomas

What is the most accurate way to ensure accurate Art Art reproduction? A team from MIT's Computer Science and Artificial Intelligence Laboratory (CSAIL) has proposed a RePaint image reproduction system that combines 3D printing and deep learning to authenticate your favorite images – regardless of the different lighting conditions or placement.

[Image: MIT CSAIL]

Traditionally, reproducing paint colors is done using a 2D printer. However, they have serious disadvantages for high-quality color reproduction because 2D printers have a fixed set of only four colors (cyan, magenta, yellow and black). Researchers have found a better way to capture the full spectrum of Degas and Dali. They used a special technique called "color rendering", which includes the use of 3D printers and 10 different transparent colors complexed in very thin layers, similar to chocolates and chocolates in the Kit-Kat bar. They have been combining their method for decades with the old technique called semi-toning, where the image is created by a multitude of small dots in color and not in continuous tones. Combining this, the team says, has better captured the shades of color.

To test RePaint, the team reproduced a series of oil paintings made by the collaborator of the author. The team found that RePaint was more than four times more precise than the most modern physical models in creating color accuracy for different works.

At this time, AI and layer layers require too much time limiting playback on the size of a business card. The system could not fully reproduce certain colors like cobalt blue because of the limited ink library. With a larger color range for work, the question of what colors to use for which images are still left. Instead of using more difficult physical access, the team has trained a deep learning model for predicting the optimum strength of different dyes. Once the system had a handle, they were fed with picture pictures and used a model to determine which colors should be used in certain areas for specific images. They can also hope to achieve better detail in order to take into account aspects such as surface texture and reflection so that they can achieve certain effects such as shiny and matte finishes.

Scientists are confident they will be able to solve all of these problems in the future, when more advanced, commercial 3D printers can help create larger images more effectively.

Source Lily (top) and reproduction (bottom). [Image: MIT CSAIL]

According to the researchers, RePaint can be used for renovation of home art, protection of wearer's original in museums, or even help companies to create prints and postcards of historical works.

"If you just play the color of an image as it looks in the gallery, it might look different in your home," says Changil Kim, one of the authors on the new system article to be presented at ACM SIGGRAPH Asia in December. "Our system works under any lighting conditions, which shows far more color reproduction capabilities than almost any other previous work."

"The value of fine art has increased rapidly in recent years, and the tendency is to move away from the public eye," said mechanical engineer Mike Foshey. "We build technology that will reverse this trend and create cheap and accurate playback that everyone can enjoy."

Published in the 3D print application

Maybe you also like:



Source link