At left is an African wood carving, and at right is a a subject's face translated into that style. All the metrics and the surfaces match the target art style.
Here, the same person's face is translated into a bronze statue. This process harnesses machine-learning techniques for style transfer, which we've seen before on the blog.
The system is not daunted by the complications of reflective surfaces and multiple light sources.
Doing this with still images is amazing enough, but what if you could animate the avatar in real time as your face moves? In this video, the still images come to life with character animation software. Link to YouTube (Start at 5:20 for the animation)
The software is called "Project Puppetron," part of "Adobe Sensei," which harnesses AI to assist graphic designers, illustrators and animators.
Adobe took pains to assure the creative professionals in the audience that tools like this won't replace artists, but instead will introduce new forms of expression. Do you agree?
----
Previously: Style Transfer
Read more in Wired: "Artificial Intelligence is Killing the Uncanny Valley and Our Grasp on Reality"
18 comments:
Wow pretty amazing.. I can see the illustration job at the Wall Street Journal taking a hit...
Did you notice that every single example in this video is using strictly en face artwork and strictly en face photo?
I smell a rat. :) AI is not really required for this; simple "morphing" technology from 30 years ago would produce a very similar result. It does not analyze the form or anything like that; it probably just distorts the artwork original until its mesh points match the corresponding mesh points of the photo. The only real use for a neural network in this would be creating optimized meshes for morphing; but any facial recognition method would probably work just as well.
So don't give up on these paints, just as yet.
It seems to me that once the process becomes procedural and the “creator” can merely enter prompts (even by voice command a la Siri, i.e. “make me a werewolf...no, bigger. More fur. Elongate the snout...cross the face with 20% bat”) the user will need good taste above drawing or painting skill. More art director/production artist than draughtsman. I give it less than 10 years.
Project Puppetron makes me shiver. Like all powerful technology, it has the potential for wonderful AND terrible results/effects. My biggest concern is that it will seduce maturing artists with an alluring shortcut to a seemingly sophisticated result. It allows you to possess something without earning it, produce something without understanding how, give yourself the feeling of a false achievement, all the while robbing you of real artistic growth.
Ask yourself, "Would I like to skip the journey and arrive instantly at my destination?" It's the allure of the teleport. And it's understandably tempting to wish to skip the exhaustion, non-linear development, and slow pace that often comes with real artistic growth. But if you make a habit of taking shortcuts, don't kid yourself, you are not an artist. The journey is the destination.
Unfortunately, the professional artist gets paid on arrival.
Jim, I was visiting a major CGI animation studio, watching a scene of two characters splashing around in shallow water. I complimented the artist on how much they must have studied refraction, reflection, and wave behavior to pull off those effects. "God no," he said. "It was all procedural. We just pressed a button and got this all on output." As John says, there will be some artistry in setting the parameters, but for some industries, artists will have to surrender much of the knowledge that made a film like Fantasia a true thing of beauty. All tools can yield true art in the hand of inspired artists, but different tools recruit different capacities.
Jim, your animation studio story is funny and sad. I imagine a person visiting a man on his deathbed. The visitor compliments the man's apparently rich life. The dying man replies, "It was all procedural. I just pressed a button and got this all on output."
Hi Jim,
The technological revolution is quickly making most of us redundant. I believe commercial artists will soon be on that list. I think there is still a demand for the "human touch" in art for home and museum display but unless the younger generations are taught a real appreciation for art, in time, human art will be replaced by AI-generated art. The 'i-generations' will use their own computer parameters to create from imagination (mechanized imagination). Artists today need to embrace the tech and keep it from becoming too cold.
Their are pros and cons to where we are headed. The primary negative aspect of technology is, for me, the separation of humanity even further from nature and into a cold realm that only regards 'big-data'.
James:
Which Fantasia movie were you referring to? The original hand-crafted in human time, or the newer composite of CG & hand-crafted?
Thanks,
Tim
bollent@wwu.edu
Tim, I haven’t seen the newer one. I was thinking especially of the Nutcracker Suite segment of the 1940 film. The lyricism of it wasn’t just because of the hand-drawn shapes and the hand painted and airbrushed cels, but also because so many of the artists were Easterners who were transplanted to L.A. with deep, longing memories of autumn leaves and ice magic.
I think the biggest problem is that when all the creating becomes "procedural" - people will no longer know the difference between actual creation and letting the computer/device/smartphone do the creating. I think we see this already with the smartphone generation, they think they are "creating" when they push a few buttons of pre-programmed effects that alter their photos, videos, etc. Yes, they are deciding between various pre-programmed alternatives - but that is not creating.
Hopefully, that actual act of creating will still be enough of a challenge and enjoyable enough for people to continue the wonderful traditions that have been part of humanity for centuries. But we may be in danger of losing them forever. Will future generations still do oil painting or watercolors or any other medium when their computers and smartphones can take their photos and turn them into any type of "painting?" I guess time will tell. Perhaps those of us that care about actual creation should not encourage or promote the technology that threatens to make it extinct.
>>>All tools can yield true art in the hand of inspired artists, but different tools recruit different capacities. <<<
Well said. Advances in the artist's tools have always had the effect of widening the success gulf between outstanding practitioners and the mediocre. Five centuries ago, there was a vast sea of second and third tier artists making livings. As the ages progressed, more money, fame and accolades went to the small percentage of those at the top and a much smaller pie was distributed among the remaining practitioners.
The average life expectancy in studios and agencies grows shorter and shorter as jejune dreams, not being realized, force the lesser talents to leave the profession and go elsewhere, teaching, opening artsy shops and other less competitive positions.
Rob, can you explain what you mean about "jejune dreams, not being realized"? Do you mean that the life cycle of entertainment companies is limited by their vision, forcing out creative talent?
Don, yes, as long as we prize original art made with physical materials, artists will have to recruit all the same skills that people have perfected over centuries of realist art. Work that inherently digital, expensive and collaborative, such as computer animation, inevitably requires technical tools and a more technical pipeline. But its really just a new art form that wasn't available to creative people in the past. I'm pretty optimistic about the potential benefits of any of these tools, especially if they mean that smaller teams of people can fulfill a vision that would have been out of reach before. Adobe's argument is that "creative intelligence" will amplify human creativity and intelligence rather than replace humans, and I think that's at least partially true.
Artificial ~ No warmth, no soul, no creative human spirit living within the works
In my mental formula for value of a work of art, the effort required to create it is one of the determinants (along with aesthetics, originality etc) In other words, I believe that we don't value things that come too easy - there has to be effort involved for us to appreciate them. If good-looking works of art start coming out at a press of a button, they would loose their appeal. Like prepackaged Instagram filters - nice, but after seeing a few, meh... We appreciate ingenuity of creation, of overcoming difficulty, coming with creative solutions for hard problems.
I think advances in artificial intelligence would probably still leave humans enjoying the fruits of their own labor, even when machines would overtake us in all endeavors, be those artistic or scientific. I think of games like chess and Go, where humans can no longer compete with machines - fine, but we can still compete against each other and enjoy it. Other sports, too. Yes, at some point a tennis playing robot could be built that would beat any human player. Would it mean that we'd no longer be interested in watching human players compete against each other?
And so with art and any intellectual labor.
What if as a result we can figure out a better way to make art. Kind of like how Frank Frazetta did it. He had a very vivid, visual imagination. He may have avoided unlearning something that is natural in everyone, when he was young.
I've learned that one of the hurdles to creating was overcoming self enforced limitations. Maybe it's a similar problem. I don't think it's being born with the right abilities.
This will push us to go beyond what we think is possible.
An anonymous commentator offers this viewpoint:
"I had to think a while before writing my very first post on a forum like this.
Coming from decades on the IT and coding side of things, I know that computer aided design is not new. AI guided decisions and expert systems have been around a long time: analyzing and mapping the decision tables that drive such systems took the lion's share of development time. Once you mapped out the logic behind an expert's decisions, it was child's play to program a system that could analyze a car's mechanical system, recommend a vacation spot, or suggest a wine paring for a dinner or cheese plate.
We've now entered the next phase, where AI will actually construct software and make business decisions about product development. I surmise that software development ("programming") will be one of the first skills to be absorbed entirely by AI. And, like the programs that automatically analyze and adjust stock funds, AI will absorb decision-making processes for running businesses and for marketing and developing products. We see glimmers of this already when ads for products we've searched for automatically pepper the websites we visit for weeks afterward. Ergo, eventually we may will see that cloud-based graphics software will self-evolve and offer product features or filters based on the types of graphic works that we view on the web or even rate on social sites.
This does not paint a picture of gloom to me. Software is another tool, not an end of art.
I cannot help but think that people will always have the need to connect to one another through art -- we need to know that out there, we can all share the experience of another human's touch. It's that thrill and awe we feel when we see a pictograph graced by the hand of an ancient and unknown artist. Likewise, sanguine pentimenti we see in a Renaissance sketch helps us understand that individual artist's creative approach. The chain of humanity reaches back into time immemorial: art makes it palpable.
I'm going to keep on drawing and painting!"
---
Thanks, Anonymous! --James G.
Post a Comment