PetaPixel

Averaged Portraits Created Using Faces Found in Popular Movies

ssbkyh_avatar copy

No, the creepy face above isn’t a still frame of He-Who-Must-Not-Be-Named taken from one of the Harry Potter films. It’s actually a composite portrait of countless faces found in the 2009 James Cameron science fiction film Avatar.

South Korean artists Shin Seung Back and Kim Yong Hun of the Seoul-based group Shinseungback Kimyonghun had a computer create the image using facial recognition software.

Here’s what they did: after loading up Avatar onto the computer, they collected every 24 frames from the film — roughly one frame for each second in the movie. They then passed the frames through the face detection program to only collect the frames that contain human faces. Finally, they lined up the faces and averaged the pixels, resulting in what you see above.

Here’s a video that shows the averaging in action:

“The composite image reflects the centric figure(s) and the visual mood of the movie,” the artists write.

Here are some of the other composite portraits they’ve created so far using other movies. See if you can determine what movie was used to make each one (answers at the end):

ssbkyh_amelie copy

ssbkyh_black_swan copy

ssbkyh_kill_bill_vol_1 copy

ssbkyh_mission_impossible copy

ssbkyh_oldboy copy

ssbkyh_taxi_driver copy

ssbkyh_the_bourne_identity copy

ssbkyh_the_matrix copy

Here are the answers in order: Amelie, Black Swan, Kill Bill (Vol 1), Mission Impossible, Oldboy, Taxi Driver, The Bourne Identity, and The Matrix.

“Portrait” by Shinseungback Kimyonghun (via Gizmodo)


Image credits: Images by Shinseungback Kimyonghun and used with permission


 
 
  • Donald

    A few months ago there were several articles on averaging simulated high multiple exposures from contributors Pat David and Bill Lytton. Bill made reference to the work of Jason Salavon http://salavon.com/work/EveryPlayboyCenterfoldDecades/grid/11/ , who on his website, paraphrasing here, writes he has abandoned it as having taken the idea as far as it can go. There is a conundrum in this human image photographic averaging process:

    Averaging multiple exposures of human faces creates average images.

    The pioneer of the process, Sir Frances Galton http://galton.org/composite.htm , experienced the same conundrum in 1879, as do this Korean group in 2013: basically average, somewhat unremarkable images, beyond a mild curiosity in the structured process and the image’s story telling collective source. It can be taken to exquisite levels as Ken Kitano does in his “Our Face” series http://www.ourface.com/english/works/ourface.html, so far the only practitioner with a finely developed aesthetic sense for this human averaging process.

    At the mild risk of spamming, I too do this work http://www.highmultiples.com , but with ordered chaos colour field aesthetic as my only driver. I discard many images that exhibit too much averageness, searching for visual alacrity in a chaotic scene, downplaying any story telling significance.

    I believe that a structured process being the raison d’etre for the end result is not good enough. The image requires more than just a process. But then, I also don’t like majority of macro photography.

  • http://www.yepi8.org/ yepi 8

    seeing faces that I think it is the ghost, and I felt a little scared

  • Island In The Sky

    CREEPY TO THE MAX!