Deep Neural Network Learns Van Gogh’s Art | Two Minute Papers #6

Deep Neural Network Learns Van Gogh’s Art | Two Minute Papers #6


Dear Fellow Scholars, this is Two Minute
Papers with Károly Zsolnai-Fehér. This paper is as fresh as it gets. As of
making of this video it has been out for only one day and I got so excited about
it that I wanted to show it to you Fellow Scholars as soon as humanly possible because you’ve got to see this. Not so long ago we have been talking
about deep neural networks, the technique that was inspired by the
human visual system. It enables computers to learn things in
a very similar way that the human would. There is a previous two minute papers
episode on this. Just click on the link in the description box if you’ve
missed it. Neural networks are by no means perfect so do not worry, don’t quit your job, you’re good, but some
applications are getting out of control. In Google Deep Mind’s case is started to
learn playing simple computer games and eventually showed a superhuman level
plays in some cases. I have run this piece of code and got some pretty sweet
results that you can check out. There is a link to it in the description box as
well. So about this paper we have here today,
what does this one do? You take photographs with your camera and you can
assign it any painting and it will apply this painting’s artistic style to it. You
can add the artistic style of Vincent van Gogh’s beautiful “Starry Night” to it and get some gorgeous results. Or, if you are looking for a bit more emotional or
may I say disturbed look you can go for Edvard Munch’s “The Scream” for some
stunning results. And of course the mandatory Picasso. So as you can see
deep neural networks are capable of amazing things and we expect even more
revolutionary works in the very near future. Thanks for watching and I’ll see you
next time!

Author: Kevin Mason

58 thoughts on “Deep Neural Network Learns Van Gogh’s Art | Two Minute Papers #6

  1. That is awesome. So interesting. Thanks for sharing. I have a huge interested in cognitive systems and machine learning, but also am a photographer/cinematographer. You hit both of my interests in one video.

  2. This is a great video! Please keep making more of these 2 minute paper videos! Subscribed! Only if authors knew how to make an "ad" for their papers 🙂

  3. I do not want to undermine creativity of "artists" on Instagram, but this will add a new dimension into "flooding" social networks with stolen art from the machines.

  4. Amazing work. What happens when you try the other way- applying the "style" of a real photograph to an impressionistic art work?

  5. ok. i just saw all your 2 minutes paper videos, now what am i supossed to doooooo??? keep em comming man!! awesome work

  6. The first implementation of the technique seems to be out! 🙂 Check this:
    http://gitxiv.com/posts/jG46ukGod8R7Rdtud/a-neural-algorithm-of-artistic-style

    Update: more are popping up! Added them to the description box.

  7. Fascinating!
    Now there NEEDS to be a competition, between artists and this neural network, with blind-judging to see who creates the better art.
    They should get some present day artists, and have them paint their version of a scene, then use this algorithm to paint the same scene, using some of those artist's past works as examples to base on. I bet the results would be interested, at the very least, and might even be useful for improving the neural network's abilities.

  8. For the handful of you that think this is somehow competition for artists, robots aren't creative the people who made the program are. On top of that the program doesn't create new style, it merely rehashes it in a new form. I think artists will use programs like this probably to great effect. But it's another tool. AI can never compete with humans when it comes to visual art, literature, music, etc. etc.

  9. I've been playing with an implementation of this algorithm. There are a few problems with current implementations but the results are still really fun. http://mlg.eng.cam.ac.uk/lloyd/blog-2015-09-01-neural-art.html and http://imgur.com/a/N7Jsk

  10. Ez nagyon tetszik, a kérdésem az, hogy ez a program(?) bárki számára hozzáférhető lesz vagy már az? Ha igen, akkor hogyan lehet ehhez hozzájutni?

  11. Speaking as an artist,I find this really interesting but could a computer as splodges of oil to a canvas the way a human does? Perhaps,but perhaps not, we shall see.

  12. Hi Károly Zsolnai-Fehér. Do you have a email address I can contact you with? I writing on behalf of a film school that is about to make their first year film project.
    – Melanie

  13. Automation threatens fork lift and truck drivers now, next maybe the creative arts and programing itself. "Learning" rather than "programming" results in "analog" look rather than "digital". Output from a process. When I fumble around with a creative thing I often enjoy the process more than the output. Hey strong, general A.I. of the future learn to "enjoy" learning this process and I grant "you" the right to vote. Wonder what artificial neural netwerks are being tasked with behind the scenes? Apply this to music? Deep feedforward neural networks will maybe even help understanding of our own wetware. In the meantime the U.N. is seeing discussions on the non-proliferation of A.I. and progressives are suggesting a universal basic income to offset massive displacement in a future of massive abundance. Wish I could wake up when it's all over . . no l'd miss out on the process.
    Don't miss out-please vote.

  14. Amazing, nice video. Tried it myself, tweaked the parameters a bit, check the results on: https://www.facebook.com/artifai/?ref=bookmarks

  15. Great video, but it makes me pensive. The photo you took to show examples is exactly the city I live in (Tübingen, Germany). I think this could be a compliment 😀

  16. Thanks for sharing. Great motivation. But I think It's better to explain more about how the paper's method or model works.

  17. Today the general public call it prizma… n now I know where it came from… Thanks a lot… 😀 BTW ..liked and subscribed.. 😉

  18. I read the paper but it's really unclear how they capture the style information in an image. Pretty much all they say is that they 'compute the correlations between the different filter responses'. Can anyone explain how it works or point me to something that explains it?

  19. Wow since then it became kind of trivial with all the apps out there that do this (Prisma for instance).

  20. Reminds of a TED talk about a neural network learning to perceive previously learned images in the contours and detail of larger images, like a formulaic way of encapsulating a kind of fractal creativity.
    However, this is more like a stylistic rendering algorithm

  21. The animals on desert picture is the most important work in my opinion… This is exactly how my etyes works in… simplified way…

  22. Hey! Thank you for constantly uploading amazing content! I am a philosopher in training and was in part inspired by your channel. Channel art came from your influence. This one might be relevant to you – peripherally: https://www.youtube.com/watch?v=QsFn4tGSCN0

  23. While impressing im not blown away to be honest, it looks like a cheap copy 'in the style of'' without any of that stylistic choice making contextual sense it remains a superficial appropriation. It is interesting though and I would love to see more developments in painting neural networks, though what I see here this approach I do not think it will achieve great results they seem focused on superficial copying of the style which doesnt elevate it much beyond a tacky photo filter letting it do multiple very different artistic styles. Id be much more interested in a in-depth study of one specific style of painting (the technique of 1 artist) but really working towards a deep understanding of that technique I think this in time can achieve very interesting results. Impressionistic paintings like that van Gogh for example are all about how to render a image of the world in a specific medium (painting) and ideas that the artist has about choices that are to be made in this rendering, I think its very applicable for this type of neural network study, would love to see a in-depth project concerning Cezanne with this, I think his still lives would be a very good subject for such a study. Not only interesting on the computer research side of it but also as opening up a whole new field of artistic/art theory research.

  24. when i see this i hope i'm still alive when AIs take over the DanceFloors and rock the human race to another dimension. 😀
    !!! PARTY ON !!!

  25. We're all so screwed…
    Social credit, replacing money with digital currency, intrusive and constant surveillance, diagnoses which change a citizen's status making them second-class citizens, rampant unemployment which also changes a person's status, a global market which has wages drop and reduce each individual's buying power, government-imposed restrictions on how much someone can earn, not being able to make money under the table, social media filtering offensive thinking and descent… not being able to escape the watchful and critical eye of our oppressors and colleagues. "OBEY!".
    https://www.youtube.com/watch?time_continue=2&v=eViswN602_k

Leave a Reply

Your email address will not be published. Required fields are marked *