Edited by: Zhang Jinhe, Song Xinyue
html On June 14, a 65-year-old Japanese artist posted a video made with Luma's Dream Machine on a social platform. In the video, he recreated his wife who had been dead for 11 years in the video, turning the cold photos into vivid images. Once the video was released, these tender feelings instantly touched the heartstrings of countless netizens. Many netizens left messages in the comment area saying they were moved to tears.
Image source: x
koya Matsuo (Matsuo Kimiya) is a 65-year-old Japanese AI artist, IT editor and musician. In order to commemorate his wife Tori-chan who passed away for 11 years, he used Luma's AI Video dream machine produced a video and posted it on social media.
Image source: x
This video has attracted the attention of many netizens and has been played more than 400,000 times. Many netizens said, "This is just like the magic in "Harry Potter"." The content of the video is as follows:
Image source: x
Some netizens left a message under this video, saying that the progress of AI technology is reflected in these places, which makes people cry.
Picture source: x
According to foreign media reports, on June 25, 2013, Matsuo Kimiya’s wife Kotori-chan unfortunately passed away due to breast cancer. After that, Mr. Matsuo also fell into a huge longing for his wife.
Whenever AI video and music generation tools came out, 65-year-old Matsuo began to study like crazy and tried to use them to "resurrection" his wife.
This time the dream machine became popular, and Mr. Matsuo was also the first to try it.
Image source: technoedge
He said in an article published in technoedge: "The purpose of using it is not for video production such as movies, but for static images, especially to make characters move. More specifically, In order to see the dynamic figure of his wife who passed away 11 years ago, "
Mr. Matsuo used film photos taken between 1978 and 1987, processed them into high-definition using remini technology, and then used luma dream machine to convert them into videos, so that he could see his wife in action. The desire to “resurrection” in the video.
Picture source: x
According to his own introduction, he uploaded a photo of his wife in high school (taken in 1978) to the dream machine, and then entered the prompt word: "a girl is talking with her friend on a moving train (moving) A girl is talking to her friend on the train)" to generate a video.
The people in the photo swayed with the vibration of the train, and the changes in shadows outside the window were also reflected.
Picture source: x
He also found several photos of his wife, first used SD to generate images, and then used Dream Machine to convert these static pictures into dynamic videos.
During the experiment, Mr. Matsuo also found that if the movement is too big, such as turning back or standing up, the face in the video will often not match the person. He speculated that this may be because the AI training data is biased towards Europeans and Americans. In order to solve this problem, he tried to use prompt words such as "... is looking at me (so-and-so is looking at me)" or "... is staring at the viewer (so-and-so is staring at the viewer)" and successfully generated Watch the video with your eyes always directed towards him.
This is a photo they took at the Tsukuba Expo in 1985.
Image source: x
The prompt word given by Matsuo was: "a girl is looking at me (a girl is looking at me)", and then the dream machine generated a video that made his heart beat every time he watched it.
Image source: x
In order to make the work more complete, he sampled his wife’s preserved voice, used the voiceflex function of suno to compose lyrics and music, and finally generated a music MV for his wife: "Can Dream Machine Dream?" ", completing a fusion of music and images that spans time and space.
Image source: x
Some netizens said: "AI has no temperature, but people who use AI do."
Daily Economic News Comprehensive Self-Disclosure Information
Daily Economic News
